deepspeed 0.18.8
Version history | Download JSON
-
PyPI https://pypi.org/project/deepspeed/ Repository https://github.com/deepspeedai/DeepSpeed -
py.typed -
Coverage
%%{init: {"pie": {"textPosition": 0.85}, "theme": "neutral", "themeVariables": {"pieStrokeWidth": "1px"}}}%% pie title "Typed" : 2989 "Any" : 37 "Untyped" : 13880 -
Typables
%%{init: {"pie": {"textPosition": 0.85}, "theme": "neutral", "themeVariables": {"pieStrokeWidth": "1px"}}}%% pie title "functions" : 3403 "classes" : 10832 "other" : 69- 1016 functions (+0 overloads)
- 2387 parameters
- 756 classes
- 4206 methods (+0 overloads)
- 6280 parameters
- 338 properties
- 4206 methods (+0 overloads)
- 555 modules
- 69 attrs
- 1016 functions (+0 overloads)
Modules
Incomplete Annotations
deepspeed (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
version_minor |
1 | 0 | 0 |
func |
set_optimizer_flags |
3 | 0 | 0 |
func |
add_config_arguments |
2 | 0 | 0 |
func |
default_inference_config |
1 | 0 | 0 |
func |
tp_model_init |
6 | 0 | 0 |
attr |
version_patch |
1 | 0 | 0 |
attr |
version_major |
1 | 0 | 0 |
func |
initialize |
14 | 7 | 0 |
func |
init_inference |
4 | 0 | 0 |
deepspeed.accelerator.abstract_accelerator (108 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedAccelerator.init |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.use_host_timers |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.device_name |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.device |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.set_device |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.current_device |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.current_device_name |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.device_count |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.synchronize |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.random |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.set_rng_state |
3 | 0 | 0 |
meth |
DeepSpeedAccelerator.get_rng_state |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.manual_seed |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.manual_seed_all |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.initial_seed |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.default_generator |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.stream |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.current_stream |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.default_stream |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.empty_cache |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.memory_allocated |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.memory_cached |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.max_memory_cached |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.memory_stats |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.memory_reserved |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.total_memory |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.available_memory |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.supported_dtypes |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_available |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.range_push |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.range_pop |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.lazy_call |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.communication_backend_name |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_triton_supported |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.create_graph |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.capture_to_graph |
4 | 0 | 0 |
meth |
DeepSpeedAccelerator.replay_graph |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.pin_memory |
3 | 0 | 0 |
meth |
DeepSpeedAccelerator.is_pinned |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.on_accelerator |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.op_builder_dir |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.create_op_builder |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.get_op_builder |
2 | 0 | 0 |
meth |
DeepSpeedAccelerator.build_extension |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.export_envs |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
DeepSpeedAccelerator.get_compile_backend |
1 | 0 | 0 |
meth |
DeepSpeedAccelerator.set_compile_backend |
2 | 0 | 0 |
prop |
DeepSpeedAccelerator.Stream |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.Event |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.ByteTensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.DoubleTensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.FloatTensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.HalfTensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.IntTensor |
1 | 0 | 0 |
prop |
DeepSpeedAccelerator.LongTensor |
1 | 0 | 0 |
deepspeed.accelerator.cpu_accelerator (111 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPU_Accelerator.init |
1 | 0 | 0 |
meth |
CPU_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
CPU_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
CPU_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
CPU_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
CPU_Accelerator.device_name |
2 | 0 | 0 |
meth |
CPU_Accelerator.device |
2 | 0 | 0 |
meth |
CPU_Accelerator.set_device |
2 | 0 | 0 |
meth |
CPU_Accelerator.current_device |
1 | 0 | 0 |
meth |
CPU_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
CPU_Accelerator.device_count |
1 | 0 | 0 |
meth |
CPU_Accelerator.synchronize |
2 | 0 | 0 |
meth |
CPU_Accelerator.random |
1 | 0 | 0 |
meth |
CPU_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
CPU_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
CPU_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
CPU_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
CPU_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
CPU_Accelerator.default_generator |
2 | 0 | 0 |
meth |
CPU_Accelerator.stream |
2 | 0 | 0 |
meth |
CPU_Accelerator.current_stream |
2 | 0 | 0 |
meth |
CPU_Accelerator.default_stream |
2 | 0 | 0 |
meth |
CPU_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
CPU_Accelerator.get_rss |
1 | 0 | 0 |
meth |
CPU_Accelerator.reset_rss |
1 | 0 | 0 |
meth |
CPU_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
CPU_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
CPU_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
CPU_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
CPU_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
CPU_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
CPU_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
CPU_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
CPU_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
CPU_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
CPU_Accelerator.total_memory |
2 | 0 | 0 |
meth |
CPU_Accelerator.available_memory |
2 | 0 | 0 |
meth |
CPU_Accelerator.is_available |
1 | 0 | 0 |
meth |
CPU_Accelerator.range_push |
2 | 0 | 0 |
meth |
CPU_Accelerator.range_pop |
1 | 0 | 0 |
meth |
CPU_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
CPU_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
CPU_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
CPU_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
CPU_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
CPU_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
CPU_Accelerator.create_graph |
1 | 0 | 0 |
meth |
CPU_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
CPU_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
CPU_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
CPU_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
CPU_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
CPU_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
CPU_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
CPU_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
CPU_Accelerator.build_extension |
1 | 0 | 0 |
meth |
CPU_Accelerator.export_envs |
1 | 0 | 0 |
meth |
CPU_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
CPU_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
CPU_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
CPU_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
CPU_Accelerator.Stream |
1 | 0 | 0 |
prop |
CPU_Accelerator.Event |
1 | 0 | 0 |
prop |
CPU_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
CPU_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
CPU_Accelerator.max_mem |
1 | 0 | 0 |
deepspeed.accelerator.cuda_accelerator (113 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CUDA_Accelerator.init |
1 | 0 | 0 |
meth |
CUDA_Accelerator._init_pynvml |
1 | 0 | 0 |
meth |
CUDA_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
CUDA_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
CUDA_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
CUDA_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
CUDA_Accelerator.device_name |
2 | 0 | 0 |
meth |
CUDA_Accelerator.communication_backend_version |
1 | 0 | 0 |
meth |
CUDA_Accelerator.device |
2 | 0 | 0 |
meth |
CUDA_Accelerator.set_device |
2 | 0 | 0 |
meth |
CUDA_Accelerator.current_device |
1 | 0 | 0 |
meth |
CUDA_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
CUDA_Accelerator.device_count |
1 | 0 | 0 |
meth |
CUDA_Accelerator.synchronize |
2 | 0 | 0 |
meth |
CUDA_Accelerator.random |
1 | 0 | 0 |
meth |
CUDA_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
CUDA_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
CUDA_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
CUDA_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
CUDA_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
CUDA_Accelerator.default_generator |
2 | 0 | 0 |
meth |
CUDA_Accelerator.stream |
2 | 0 | 0 |
meth |
CUDA_Accelerator.current_stream |
2 | 0 | 0 |
meth |
CUDA_Accelerator.default_stream |
2 | 0 | 0 |
meth |
CUDA_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
CUDA_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
CUDA_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
CUDA_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
CUDA_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
CUDA_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
CUDA_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
CUDA_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
CUDA_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
CUDA_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
CUDA_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
CUDA_Accelerator.total_memory |
2 | 0 | 0 |
meth |
CUDA_Accelerator._get_nvml_gpu_id |
2 | 0 | 0 |
meth |
CUDA_Accelerator.available_memory |
2 | 0 | 0 |
meth |
CUDA_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
CUDA_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
CUDA_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
CUDA_Accelerator.is_available |
1 | 0 | 0 |
meth |
CUDA_Accelerator.range_push |
2 | 0 | 0 |
meth |
CUDA_Accelerator.range_pop |
1 | 0 | 0 |
meth |
CUDA_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
CUDA_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
CUDA_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
CUDA_Accelerator.create_graph |
1 | 0 | 0 |
meth |
CUDA_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
CUDA_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
CUDA_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
CUDA_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
CUDA_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
CUDA_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
CUDA_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
CUDA_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
CUDA_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
CUDA_Accelerator.build_extension |
1 | 0 | 0 |
meth |
CUDA_Accelerator.export_envs |
1 | 0 | 0 |
meth |
CUDA_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
CUDA_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
CUDA_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
CUDA_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
CUDA_Accelerator.Stream |
1 | 0 | 0 |
prop |
CUDA_Accelerator.Event |
1 | 0 | 0 |
prop |
CUDA_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
CUDA_Accelerator.LongTensor |
1 | 0 | 0 |
deepspeed.accelerator.hpu_accelerator (112 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HPU_Accelerator.init |
1 | 0 | 0 |
meth |
HPU_Accelerator.apply_hpu_workarounds |
1 | 0 | 0 |
meth |
HPU_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
HPU_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
HPU_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
HPU_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
HPU_Accelerator.device_name |
2 | 0 | 0 |
meth |
HPU_Accelerator.device |
2 | 0 | 0 |
meth |
HPU_Accelerator.set_device |
2 | 0 | 0 |
meth |
HPU_Accelerator.current_device |
1 | 0 | 0 |
meth |
HPU_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
HPU_Accelerator.device_count |
1 | 0 | 0 |
meth |
HPU_Accelerator.synchronize |
2 | 0 | 0 |
meth |
HPU_Accelerator.random |
1 | 0 | 0 |
meth |
HPU_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
HPU_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
HPU_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
HPU_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
HPU_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
HPU_Accelerator.default_generator |
2 | 0 | 0 |
meth |
HPU_Accelerator.stream |
2 | 0 | 0 |
meth |
HPU_Accelerator.current_stream |
2 | 0 | 0 |
meth |
HPU_Accelerator.default_stream |
2 | 0 | 0 |
meth |
HPU_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
HPU_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
HPU_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
HPU_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
HPU_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
HPU_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
HPU_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
HPU_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
HPU_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
HPU_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
HPU_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
HPU_Accelerator.total_memory |
2 | 0 | 0 |
meth |
HPU_Accelerator.available_memory |
2 | 0 | 0 |
meth |
HPU_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
HPU_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
HPU_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
HPU_Accelerator.is_available |
1 | 0 | 0 |
meth |
HPU_Accelerator.range_push |
2 | 0 | 0 |
meth |
HPU_Accelerator.range_pop |
1 | 0 | 0 |
meth |
HPU_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
HPU_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
HPU_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
HPU_Accelerator.create_graph |
1 | 0 | 0 |
meth |
HPU_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
HPU_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
HPU_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
HPU_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
HPU_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
HPU_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
HPU_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
HPU_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
HPU_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
HPU_Accelerator.build_extension |
1 | 0 | 0 |
meth |
HPU_Accelerator.export_envs |
1 | 0 | 0 |
meth |
HPU_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
HPU_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
HPU_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
HPU_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
HPU_Accelerator.Stream |
1 | 0 | 0 |
prop |
HPU_Accelerator.Event |
1 | 0 | 0 |
prop |
HPU_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
HPU_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
HPU_Accelerator.fp16_supported |
1 | 0 | 0 |
attr |
HPU_Accelerator.hpu |
1 | 0 | 0 |
deepspeed.accelerator.mlu_accelerator (110 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MLU_Accelerator.init |
1 | 0 | 0 |
meth |
MLU_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
MLU_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
MLU_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
MLU_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
MLU_Accelerator.device_name |
2 | 0 | 0 |
meth |
MLU_Accelerator.device |
2 | 0 | 0 |
meth |
MLU_Accelerator.set_device |
2 | 0 | 0 |
meth |
MLU_Accelerator.current_device |
1 | 0 | 0 |
meth |
MLU_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
MLU_Accelerator.device_count |
1 | 0 | 0 |
meth |
MLU_Accelerator.synchronize |
2 | 0 | 0 |
meth |
MLU_Accelerator.random |
1 | 0 | 0 |
meth |
MLU_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
MLU_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
MLU_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
MLU_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
MLU_Accelerator.initial_seed |
2 | 0 | 0 |
meth |
MLU_Accelerator.default_generator |
2 | 0 | 0 |
meth |
MLU_Accelerator.stream |
2 | 0 | 0 |
meth |
MLU_Accelerator.current_stream |
2 | 0 | 0 |
meth |
MLU_Accelerator.default_stream |
2 | 0 | 0 |
meth |
MLU_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
MLU_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
MLU_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
MLU_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
MLU_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
MLU_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
MLU_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
MLU_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
MLU_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
MLU_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
MLU_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
MLU_Accelerator.total_memory |
2 | 0 | 0 |
meth |
MLU_Accelerator.available_memory |
2 | 0 | 0 |
meth |
MLU_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
MLU_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
MLU_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
MLU_Accelerator.is_available |
1 | 0 | 0 |
meth |
MLU_Accelerator.range_push |
2 | 0 | 0 |
meth |
MLU_Accelerator.range_pop |
1 | 0 | 0 |
meth |
MLU_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
MLU_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
MLU_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
MLU_Accelerator.create_graph |
1 | 0 | 0 |
meth |
MLU_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
MLU_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
MLU_Accelerator.pin_memory |
2 | 0 | 0 |
meth |
MLU_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
MLU_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
MLU_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
MLU_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
MLU_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
MLU_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
MLU_Accelerator.build_extension |
1 | 0 | 0 |
meth |
MLU_Accelerator.export_envs |
1 | 0 | 0 |
meth |
MLU_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
MLU_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
MLU_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
MLU_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
MLU_Accelerator.Stream |
1 | 0 | 0 |
prop |
MLU_Accelerator.Event |
1 | 0 | 0 |
prop |
MLU_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
MLU_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
MLU_Accelerator.class_dict |
1 | 0 | 0 |
deepspeed.accelerator.mps_accelerator (111 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MPS_Accelerator.init |
1 | 0 | 0 |
meth |
MPS_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
MPS_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
MPS_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
MPS_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
MPS_Accelerator.device_name |
2 | 0 | 0 |
meth |
MPS_Accelerator.device |
2 | 0 | 0 |
meth |
MPS_Accelerator.set_device |
2 | 0 | 0 |
meth |
MPS_Accelerator.current_device |
1 | 0 | 0 |
meth |
MPS_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
MPS_Accelerator.device_count |
1 | 0 | 0 |
meth |
MPS_Accelerator.synchronize |
2 | 0 | 0 |
meth |
MPS_Accelerator.random |
1 | 0 | 0 |
meth |
MPS_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
MPS_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
MPS_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
MPS_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
MPS_Accelerator.seed |
1 | 0 | 0 |
meth |
MPS_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
MPS_Accelerator.default_generator |
2 | 0 | 0 |
meth |
MPS_Accelerator.stream |
2 | 0 | 0 |
meth |
MPS_Accelerator.current_stream |
2 | 0 | 0 |
meth |
MPS_Accelerator.default_stream |
2 | 0 | 0 |
meth |
MPS_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
MPS_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
MPS_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
MPS_Accelerator.set_per_process_memory_fraction |
2 | 0 | 0 |
meth |
MPS_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
MPS_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
MPS_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
MPS_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
MPS_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
MPS_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
MPS_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
MPS_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
MPS_Accelerator.total_memory |
2 | 0 | 0 |
meth |
MPS_Accelerator.available_memory |
2 | 0 | 0 |
meth |
MPS_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
MPS_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
MPS_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
MPS_Accelerator.is_available |
1 | 0 | 0 |
meth |
MPS_Accelerator.range_push |
2 | 0 | 0 |
meth |
MPS_Accelerator.range_pop |
1 | 0 | 0 |
meth |
MPS_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
MPS_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
MPS_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
MPS_Accelerator.create_graph |
1 | 0 | 0 |
meth |
MPS_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
MPS_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
MPS_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
MPS_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
MPS_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
MPS_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
MPS_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
MPS_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
MPS_Accelerator.build_extension |
1 | 0 | 0 |
meth |
MPS_Accelerator.export_envs |
1 | 0 | 0 |
meth |
MPS_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
MPS_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
MPS_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
MPS_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
MPS_Accelerator.Stream |
1 | 0 | 0 |
prop |
MPS_Accelerator.Event |
1 | 0 | 0 |
prop |
MPS_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
MPS_Accelerator.LongTensor |
1 | 0 | 0 |
deepspeed.accelerator.npu_accelerator (110 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NPU_Accelerator.init |
1 | 0 | 0 |
meth |
NPU_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
NPU_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
NPU_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
NPU_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
NPU_Accelerator.device_name |
2 | 0 | 0 |
meth |
NPU_Accelerator.device |
2 | 0 | 0 |
meth |
NPU_Accelerator.set_device |
2 | 0 | 0 |
meth |
NPU_Accelerator.current_device |
1 | 0 | 0 |
meth |
NPU_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
NPU_Accelerator.device_count |
1 | 0 | 0 |
meth |
NPU_Accelerator.synchronize |
2 | 0 | 0 |
meth |
NPU_Accelerator.random |
1 | 0 | 0 |
meth |
NPU_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
NPU_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
NPU_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
NPU_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
NPU_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
NPU_Accelerator.default_generator |
2 | 0 | 0 |
meth |
NPU_Accelerator.stream |
2 | 0 | 0 |
meth |
NPU_Accelerator.current_stream |
2 | 0 | 0 |
meth |
NPU_Accelerator.default_stream |
2 | 0 | 0 |
meth |
NPU_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
NPU_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
NPU_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
NPU_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
NPU_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
NPU_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
NPU_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
NPU_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
NPU_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
NPU_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
NPU_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
NPU_Accelerator.total_memory |
2 | 0 | 0 |
meth |
NPU_Accelerator.available_memory |
2 | 0 | 0 |
meth |
NPU_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
NPU_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
NPU_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
NPU_Accelerator.is_available |
1 | 0 | 0 |
meth |
NPU_Accelerator.range_push |
2 | 0 | 0 |
meth |
NPU_Accelerator.range_pop |
1 | 0 | 0 |
meth |
NPU_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
NPU_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
NPU_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
NPU_Accelerator.create_graph |
1 | 0 | 0 |
meth |
NPU_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
NPU_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
NPU_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
NPU_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
NPU_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
NPU_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
NPU_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
NPU_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
NPU_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
NPU_Accelerator.build_extension |
1 | 0 | 0 |
meth |
NPU_Accelerator.export_envs |
1 | 0 | 0 |
meth |
NPU_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
NPU_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
NPU_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
NPU_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
NPU_Accelerator.Stream |
1 | 0 | 0 |
prop |
NPU_Accelerator.Event |
1 | 0 | 0 |
prop |
NPU_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
NPU_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
NPU_Accelerator.class_dict |
1 | 0 | 0 |
deepspeed.accelerator.real_accelerator (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_accelerator |
1 | 0 | 0 |
func |
set_accelerator |
2 | 0 | 0 |
func |
is_current_accelerator_supported |
1 | 0 | 0 |
deepspeed.accelerator.sdaa_accelerator (110 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SDAA_Accelerator.init |
1 | 0 | 0 |
meth |
SDAA_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
SDAA_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
SDAA_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
SDAA_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
SDAA_Accelerator.device_name |
2 | 0 | 0 |
meth |
SDAA_Accelerator.device |
2 | 0 | 0 |
meth |
SDAA_Accelerator.set_device |
2 | 0 | 0 |
meth |
SDAA_Accelerator.current_device |
1 | 0 | 0 |
meth |
SDAA_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
SDAA_Accelerator.device_count |
1 | 0 | 0 |
meth |
SDAA_Accelerator.synchronize |
2 | 0 | 0 |
meth |
SDAA_Accelerator.random |
1 | 0 | 0 |
meth |
SDAA_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
SDAA_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
SDAA_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
SDAA_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
SDAA_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
SDAA_Accelerator.default_generator |
2 | 0 | 0 |
meth |
SDAA_Accelerator.stream |
2 | 0 | 0 |
meth |
SDAA_Accelerator.current_stream |
2 | 0 | 0 |
meth |
SDAA_Accelerator.default_stream |
2 | 0 | 0 |
meth |
SDAA_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
SDAA_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
SDAA_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
SDAA_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
SDAA_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
SDAA_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
SDAA_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
SDAA_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
SDAA_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
SDAA_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
SDAA_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
SDAA_Accelerator.total_memory |
2 | 0 | 0 |
meth |
SDAA_Accelerator.available_memory |
2 | 0 | 0 |
meth |
SDAA_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
SDAA_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
SDAA_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
SDAA_Accelerator.is_available |
1 | 0 | 0 |
meth |
SDAA_Accelerator.range_push |
2 | 0 | 0 |
meth |
SDAA_Accelerator.range_pop |
1 | 0 | 0 |
meth |
SDAA_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
SDAA_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
SDAA_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
SDAA_Accelerator.create_graph |
1 | 0 | 0 |
meth |
SDAA_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
SDAA_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
SDAA_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
SDAA_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
SDAA_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
SDAA_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
SDAA_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
SDAA_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
SDAA_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
SDAA_Accelerator.build_extension |
1 | 0 | 0 |
meth |
SDAA_Accelerator.export_envs |
1 | 0 | 0 |
meth |
SDAA_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
SDAA_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
SDAA_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
SDAA_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
SDAA_Accelerator.Stream |
1 | 0 | 0 |
prop |
SDAA_Accelerator.Event |
1 | 0 | 0 |
prop |
SDAA_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
SDAA_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
SDAA_Accelerator.class_dict |
1 | 0 | 0 |
deepspeed.accelerator.xpu_accelerator (111 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
XPU_Accelerator.init |
1 | 0 | 0 |
meth |
XPU_Accelerator.is_synchronized_device |
1 | 0 | 0 |
meth |
XPU_Accelerator.use_host_timers |
1 | 0 | 0 |
meth |
XPU_Accelerator.resolves_data_dependency |
1 | 0 | 0 |
meth |
XPU_Accelerator.handles_memory_backpressure |
1 | 0 | 0 |
meth |
XPU_Accelerator.device_name |
2 | 0 | 0 |
meth |
XPU_Accelerator.device |
2 | 0 | 0 |
meth |
XPU_Accelerator.set_device |
2 | 0 | 0 |
meth |
XPU_Accelerator.current_device |
1 | 0 | 0 |
meth |
XPU_Accelerator.current_device_name |
1 | 0 | 0 |
meth |
XPU_Accelerator.device_count |
1 | 0 | 0 |
meth |
XPU_Accelerator.synchronize |
2 | 0 | 0 |
meth |
XPU_Accelerator.random |
1 | 0 | 0 |
meth |
XPU_Accelerator.set_rng_state |
3 | 0 | 0 |
meth |
XPU_Accelerator.get_rng_state |
2 | 0 | 0 |
meth |
XPU_Accelerator.manual_seed |
2 | 0 | 0 |
meth |
XPU_Accelerator.manual_seed_all |
2 | 0 | 0 |
meth |
XPU_Accelerator.initial_seed |
1 | 0 | 0 |
meth |
XPU_Accelerator.default_generator |
2 | 0 | 0 |
meth |
XPU_Accelerator.stream |
2 | 0 | 0 |
meth |
XPU_Accelerator.current_stream |
2 | 0 | 0 |
meth |
XPU_Accelerator.default_stream |
2 | 0 | 0 |
meth |
XPU_Accelerator.empty_cache |
1 | 0 | 0 |
meth |
XPU_Accelerator.memory_allocated |
2 | 0 | 0 |
meth |
XPU_Accelerator.max_memory_allocated |
2 | 0 | 0 |
meth |
XPU_Accelerator.reset_max_memory_allocated |
2 | 0 | 0 |
meth |
XPU_Accelerator.memory_cached |
2 | 0 | 0 |
meth |
XPU_Accelerator.max_memory_cached |
2 | 0 | 0 |
meth |
XPU_Accelerator.reset_max_memory_cached |
2 | 0 | 0 |
meth |
XPU_Accelerator.memory_stats |
2 | 0 | 0 |
meth |
XPU_Accelerator.reset_peak_memory_stats |
2 | 0 | 0 |
meth |
XPU_Accelerator.memory_reserved |
2 | 0 | 0 |
meth |
XPU_Accelerator.max_memory_reserved |
2 | 0 | 0 |
meth |
XPU_Accelerator.total_memory |
2 | 0 | 0 |
meth |
XPU_Accelerator.available_memory |
2 | 0 | 0 |
meth |
XPU_Accelerator.is_available |
1 | 0 | 0 |
meth |
XPU_Accelerator.range_push |
2 | 0 | 0 |
meth |
XPU_Accelerator.range_pop |
1 | 0 | 0 |
meth |
XPU_Accelerator.lazy_call |
2 | 0 | 0 |
meth |
XPU_Accelerator.communication_backend_name |
1 | 0 | 0 |
meth |
XPU_Accelerator.is_triton_supported |
1 | 0 | 0 |
meth |
XPU_Accelerator.create_graph |
1 | 0 | 0 |
meth |
XPU_Accelerator.capture_to_graph |
4 | 0 | 0 |
meth |
XPU_Accelerator.replay_graph |
2 | 0 | 0 |
meth |
XPU_Accelerator.is_bf16_supported |
1 | 0 | 0 |
meth |
XPU_Accelerator.is_fp16_supported |
1 | 0 | 0 |
meth |
XPU_Accelerator.supported_dtypes |
1 | 0 | 0 |
meth |
XPU_Accelerator.pin_memory |
3 | 0 | 0 |
meth |
XPU_Accelerator.is_pinned |
2 | 0 | 0 |
meth |
XPU_Accelerator.op_builder_dir |
1 | 0 | 0 |
meth |
XPU_Accelerator.on_accelerator |
2 | 0 | 0 |
meth |
XPU_Accelerator._lazy_init_class_dict |
1 | 0 | 0 |
meth |
XPU_Accelerator.create_op_builder |
2 | 0 | 0 |
meth |
XPU_Accelerator.get_op_builder |
2 | 0 | 0 |
meth |
XPU_Accelerator.build_extension |
1 | 0 | 0 |
meth |
XPU_Accelerator.export_envs |
1 | 0 | 0 |
meth |
XPU_Accelerator.visible_devices_envs |
1 | 0 | 0 |
meth |
XPU_Accelerator.set_visible_devices_envs |
3 | 0 | 0 |
meth |
XPU_Accelerator.get_compile_backend |
1 | 0 | 0 |
meth |
XPU_Accelerator.set_compile_backend |
2 | 0 | 0 |
prop |
XPU_Accelerator.Stream |
1 | 0 | 0 |
prop |
XPU_Accelerator.Event |
1 | 0 | 0 |
prop |
XPU_Accelerator.BFloat16Tensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.ByteTensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.DoubleTensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.FloatTensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.HalfTensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.IntTensor |
1 | 0 | 0 |
prop |
XPU_Accelerator.LongTensor |
1 | 0 | 0 |
attr |
XPU_Accelerator.aligned_tensors |
1 | 0 | 0 |
attr |
XPU_Accelerator.class_dict |
1 | 0 | 0 |
deepspeed.autotuning.autotuner (76 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Autotuner.init |
3 | 0 | 0 |
meth |
Autotuner.print_tuning_results |
1 | 0 | 0 |
meth |
Autotuner._get_user_config |
2 | 0 | 0 |
meth |
Autotuner._get_resource_manager |
2 | 0 | 0 |
meth |
Autotuner._get_exp_resources |
2 | 0 | 0 |
meth |
Autotuner.metric |
1 | 0 | 0 |
meth |
Autotuner.fast_enabled |
1 | 0 | 0 |
meth |
Autotuner.max_train_batch_size |
1 | 0 | 0 |
meth |
Autotuner.mp_size |
1 | 0 | 0 |
meth |
Autotuner.max_train_micro_batch_size_per_gpu |
1 | 0 | 0 |
meth |
Autotuner.min_train_micro_batch_size_per_gpu |
1 | 0 | 0 |
meth |
Autotuner.num_tuning_micro_batch_sizes |
1 | 0 | 0 |
meth |
Autotuner.fp16_enabled |
1 | 0 | 0 |
meth |
Autotuner.get_gpu_memory_info |
1 | 0 | 0 |
meth |
Autotuner.get_activation_memory_per_gpu |
1 | 0 | 0 |
meth |
Autotuner.get_instantiation_memory_required_per_gpu |
2 | 0 | 0 |
meth |
Autotuner._generate_experiments |
3 | 0 | 0 |
meth |
Autotuner.tune |
1 | 0 | 0 |
meth |
Autotuner.tune_space |
5 | 0 | 0 |
meth |
Autotuner.get_plateau_mbs |
2 | 0 | 0 |
meth |
Autotuner.get_model_num_params |
1 | 0 | 0 |
meth |
Autotuner.model_info_profile_run |
1 | 0 | 0 |
meth |
Autotuner.update_records |
5 | 0 | 0 |
meth |
Autotuner.get_best_space_record |
2 | 0 | 0 |
meth |
Autotuner.get_best_space_records |
1 | 0 | 0 |
meth |
Autotuner.run_tuning_micro_batch_sizes |
6 | 0 | 0 |
meth |
Autotuner.get_min_max_micro_batch_size |
4 | 0 | 0 |
meth |
Autotuner.get_gas_from_user_config |
1 | 0 | 0 |
meth |
Autotuner.get_val_from_user_args |
2 | 0 | 0 |
meth |
Autotuner.get_tuning_micro_batch_size_list |
4 | 0 | 0 |
meth |
Autotuner.run_ds_config |
3 | 0 | 0 |
meth |
Autotuner.write_optimal_config |
1 | 0 | 0 |
meth |
Autotuner.run_after_tuning |
1 | 0 | 0 |
attr |
Autotuner.args |
1 | 0 | 0 |
attr |
Autotuner.selected_exp_dir |
1 | 0 | 0 |
attr |
Autotuner.user_config |
1 | 0 | 0 |
attr |
Autotuner.autotuning_config |
1 | 0 | 0 |
attr |
Autotuner.exps_dir |
1 | 0 | 0 |
attr |
Autotuner.results_dir |
1 | 0 | 0 |
attr |
Autotuner.rm |
1 | 0 | 0 |
attr |
Autotuner.records |
1 | 0 | 0 |
attr |
Autotuner.optimal_cmd |
1 | 0 | 0 |
attr |
Autotuner.optimal_ds_config |
1 | 0 | 0 |
attr |
Autotuner.mlflow_parent_id |
1 | 0 | 0 |
deepspeed.autotuning.config (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedAutotuningConfig.init |
2 | 0 | 0 |
meth |
DeepSpeedAutotuningConfig._initialize |
2 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.enabled |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.start_step |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.end_step |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.metric_path |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.arg_mappings |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.metric |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.model_info |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.results_dir |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.exps_dir |
1 | 0 | 0 |
attr |
DeepSpeedAutotuningConfig.overwrite |
1 | 0 | 0 |
func |
get_model_info_config |
2 | 0 | 0 |
func |
get_default_model_info_config |
1 | 0 | 0 |
deepspeed.autotuning.constants (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
DEFAULT_TEMPLATE_PATH_ZERO_2 |
1 | 0 | 0 |
attr |
DEFAULT_TEMPLATE_PATH_ZERO_3 |
1 | 0 | 0 |
attr |
DEFAULT_TEMPLATE_PATH_ZERO_1 |
1 | 0 | 0 |
attr |
DEFAULT_TEMPLATE_PATH_ZERO_0 |
1 | 0 | 0 |
deepspeed.autotuning.scheduler (57 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
run_experiment |
5 | 1 | 0 |
func |
get_user |
1 | 0 | 0 |
meth |
Reservation.init |
3 | 0 | 0 |
meth |
Reservation.restore_slots |
1 | 0 | 0 |
meth |
Reservation.desc |
1 | 0 | 0 |
attr |
Reservation.node |
1 | 0 | 0 |
attr |
Reservation.slots |
1 | 0 | 0 |
meth |
ResourceManager.init |
7 | 0 | 0 |
meth |
ResourceManager.schedule_experiments |
2 | 0 | 0 |
meth |
ResourceManager.run_job |
3 | 1 | 0 |
meth |
ResourceManager.experiment_check |
2 | 0 | 0 |
meth |
ResourceManager.resource_request |
2 | 0 | 0 |
meth |
ResourceManager.status |
1 | 0 | 0 |
meth |
ResourceManager.run |
1 | 0 | 0 |
meth |
ResourceManager.save_exp_results_to_database |
4 | 0 | 0 |
meth |
ResourceManager.parse_results |
2 | 0 | 0 |
meth |
ResourceManager.clear |
1 | 0 | 0 |
attr |
ResourceManager.results_dir |
1 | 0 | 0 |
attr |
ResourceManager.exps_dir |
1 | 0 | 0 |
attr |
ResourceManager.nodes |
1 | 0 | 0 |
attr |
ResourceManager.num_gpus_per_node |
1 | 0 | 0 |
attr |
ResourceManager.experiment_queue |
1 | 0 | 0 |
attr |
ResourceManager.running_experiments |
1 | 0 | 0 |
attr |
ResourceManager.finished_experiments |
1 | 0 | 0 |
attr |
ResourceManager.experiment_count |
1 | 0 | 0 |
attr |
ResourceManager.exp_paths |
1 | 0 | 0 |
attr |
ResourceManager.args |
1 | 0 | 0 |
attr |
ResourceManager.arg_mappings |
1 | 0 | 0 |
func |
get_job_id |
1 | 0 | 0 |
meth |
Node.init |
3 | 0 | 0 |
meth |
Node.restore_slots |
2 | 1 | 0 |
attr |
Node.host |
1 | 0 | 0 |
attr |
Node.max_slots |
1 | 0 | 0 |
attr |
Node.idle_slots |
1 | 0 | 0 |
func |
clean_up |
3 | 1 | 0 |
deepspeed.autotuning.tuner.base_tuner (18 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BaseTuner.init |
4 | 0 | 0 |
meth |
BaseTuner.has_next |
1 | 0 | 0 |
meth |
BaseTuner.next_batch |
2 | 0 | 0 |
meth |
BaseTuner.update |
1 | 0 | 0 |
meth |
BaseTuner.tune |
4 | 0 | 0 |
attr |
BaseTuner.all_exps |
1 | 0 | 0 |
attr |
BaseTuner.rm |
1 | 0 | 0 |
attr |
BaseTuner.best_iter |
1 | 0 | 0 |
attr |
BaseTuner.best_exp |
1 | 0 | 0 |
attr |
BaseTuner.best_metric_val |
1 | 0 | 0 |
attr |
BaseTuner.metric |
1 | 0 | 0 |
deepspeed.autotuning.tuner.cost_model (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
XGBoostCostModel.init |
5 | 0 | 0 |
meth |
XGBoostCostModel.fit |
3 | 0 | 0 |
meth |
XGBoostCostModel.predict |
2 | 0 | 0 |
attr |
XGBoostCostModel.loss_type |
1 | 0 | 0 |
attr |
XGBoostCostModel.xgb_params |
1 | 0 | 0 |
deepspeed.autotuning.tuner.index_based_tuner (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GridSearchTuner.init |
4 | 1 | 0 |
meth |
GridSearchTuner.next_batch |
2 | 0 | 0 |
meth |
RandomTuner.init |
4 | 1 | 0 |
meth |
RandomTuner.next_batch |
2 | 0 | 0 |
deepspeed.autotuning.tuner.model_based_tuner (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ModelBasedTuner.init |
5 | 1 | 0 |
meth |
ModelBasedTuner.find_estimated_top_configs |
1 | 0 | 0 |
meth |
ModelBasedTuner.next_batch |
2 | 0 | 0 |
meth |
ModelBasedTuner.has_next |
1 | 0 | 0 |
meth |
ModelBasedTuner.update |
1 | 0 | 0 |
attr |
ModelBasedTuner.tuning_space |
1 | 0 | 0 |
attr |
ModelBasedTuner.best_iter |
1 | 0 | 0 |
attr |
ModelBasedTuner.all_configs |
1 | 0 | 0 |
attr |
ModelBasedTuner.num_all_configs |
1 | 0 | 0 |
attr |
ModelBasedTuner.dims |
1 | 0 | 0 |
attr |
ModelBasedTuner.visited |
1 | 0 | 0 |
attr |
ModelBasedTuner.trials |
1 | 0 | 0 |
attr |
ModelBasedTuner.trial_pt |
1 | 0 | 0 |
attr |
ModelBasedTuner.cost_model |
1 | 0 | 0 |
attr |
ModelBasedTuner.evaluated_configs |
1 | 0 | 0 |
attr |
ModelBasedTuner.evaluated_perf |
1 | 0 | 0 |
attr |
ModelBasedTuner.train_ct |
1 | 0 | 0 |
attr |
ModelBasedTuner.random_exploration_ratio |
1 | 0 | 0 |
deepspeed.autotuning.tuner.utils (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
dict_to_feature |
4 | 0 | 0 |
func |
gen_combinations |
2 | 1 | 0 |
func |
dict_to_dims |
2 | 0 | 0 |
func |
feature_to_index |
3 | 0 | 0 |
func |
index_to_feature |
3 | 0 | 0 |
func |
flatten |
4 | 0 | 0 |
deepspeed.autotuning.utils (57 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
remove_dupe_dicts |
2 | 0 | 0 |
func |
get_tuning_keys |
2 | 1 | 0 |
func |
write_experiments |
3 | 2 | 0 |
func |
get_first_config |
2 | 1 | 0 |
func |
get_val_by_key |
3 | 1 | 0 |
func |
combine_dict |
3 | 0 | 0 |
func |
get_list |
2 | 0 | 0 |
func |
find_replace |
3 | 0 | 0 |
func |
find_replace_str |
3 | 0 | 0 |
func |
get_all_configs |
3 | 1 | 0 |
func |
search_error |
2 | 0 | 0 |
func |
fetch_hostfile |
2 | 0 | 0 |
func |
prune_config |
3 | 0 | 0 |
func |
validate_ds_config |
2 | 1 | 0 |
func |
was_interrupted |
2 | 0 | 0 |
func |
memory_to_string |
5 | 0 | 0 |
func |
prune_configs |
3 | 0 | 0 |
func |
number_to_string |
5 | 0 | 0 |
func |
del_if_exists |
3 | 0 | 0 |
func |
replace_dict |
4 | 0 | 0 |
func |
canonical_name |
5 | 1 | 0 |
func |
set_val_by_key |
4 | 1 | 0 |
deepspeed.checkpoint.deepspeed_checkpoint (63 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedCheckpoint.init |
6 | 0 | 0 |
meth |
DeepSpeedCheckpoint.is_change_tp_degree |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.is_change_pp_degree |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.is_change_dp_degree |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.show_2d_mapping |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.show_tp_embedding_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.show_tp_final_norm_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.show_pp_transformer_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.show_transformer_file_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._build_global_state |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.get_zero_checkpoint_state |
5 | 2 | 0 |
meth |
DeepSpeedCheckpoint.get_zero_files |
4 | 1 | 0 |
meth |
DeepSpeedCheckpoint.get_embedding_layer_id |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.get_final_norm_layer_id |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.get_iteration |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._get_checkpoint_value |
2 | 0 | 0 |
meth |
DeepSpeedCheckpoint.get_args |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.get_checkpoint_info |
2 | 0 | 0 |
meth |
DeepSpeedCheckpoint._build_tp_other_layer_map |
2 | 1 | 0 |
meth |
DeepSpeedCheckpoint._build_pp_transformer_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._dump_mapping |
3 | 0 | 0 |
meth |
DeepSpeedCheckpoint._build_transformer_file_map |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._sanity_check |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint.validate_files |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._get_layer_keys |
1 | 0 | 0 |
meth |
DeepSpeedCheckpoint._merge_state_dicts |
2 | 0 | 0 |
meth |
DeepSpeedCheckpoint._validate_folder |
3 | 0 | 0 |
attr |
DeepSpeedCheckpoint.final_layer_norm_idx |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.dir |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.zero_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.file_list |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.layer_files |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.mp_rank_files |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.layer_keys |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.layer_count |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.tp_degree |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.pp_degree |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.dp_degree |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.original_world_size |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.world_size |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.old_2d_map |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.new_2d_map |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.global_state |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.pp_to_transformer_map |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.transformer_file_map |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.tp_to_embedding_map |
1 | 0 | 0 |
attr |
DeepSpeedCheckpoint.tp_to_final_norm_map |
1 | 0 | 0 |
deepspeed.checkpoint.ds_to_universal (40 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
merge_zero3_slices |
5 | 0 | 0 |
func |
atoi |
2 | 0 | 0 |
func |
dump_param_fragment |
9 | 0 | 0 |
func |
merge_tp_slices |
6 | 0 | 0 |
func |
extract_zero_shards |
4 | 0 | 0 |
func |
extract_zero_shards_stage3 |
6 | 0 | 0 |
attr |
args |
1 | 0 | 0 |
func |
dp_index_to_str |
2 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
func |
main |
2 | 0 | 0 |
func |
natural_keys |
2 | 0 | 0 |
deepspeed.checkpoint.reshape_3d_utils (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
unflatten_dp_dimension |
3 | 0 | 0 |
meth |
model_3d_desc.init |
4 | 0 | 0 |
meth |
model_3d_desc.reshape |
3 | 0 | 0 |
meth |
model_3d_desc.get_desc |
1 | 0 | 0 |
meth |
model_3d_desc.world_size |
1 | 0 | 0 |
meth |
model_3d_desc.is_valid |
4 | 0 | 0 |
meth |
model_3d_desc.can_reshape |
2 | 0 | 0 |
attr |
model_3d_desc.pp_degree |
1 | 0 | 0 |
attr |
model_3d_desc.tp_degree |
1 | 0 | 0 |
attr |
model_3d_desc.dp_degree |
1 | 0 | 0 |
func |
get_model_3d_descriptor |
2 | 0 | 0 |
func |
flatten_dp_dimension |
4 | 0 | 0 |
deepspeed.checkpoint.reshape_meg_2d (36 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
meg_2d_parallel_map.init |
3 | 0 | 0 |
meth |
meg_2d_parallel_map.simple_init |
1 | 0 | 0 |
meth |
meg_2d_parallel_map.add_data |
4 | 0 | 0 |
meth |
meg_2d_parallel_map.get_data |
3 | 0 | 0 |
meth |
meg_2d_parallel_map.print_data |
2 | 0 | 0 |
meth |
meg_2d_parallel_map._validate_indices |
3 | 0 | 0 |
meth |
meg_2d_parallel_map._make_key |
3 | 0 | 0 |
attr |
meg_2d_parallel_map.pp_degree |
1 | 0 | 0 |
attr |
meg_2d_parallel_map.tp_degree |
1 | 0 | 0 |
attr |
meg_2d_parallel_map.map |
1 | 0 | 0 |
func |
reshape |
3 | 0 | 0 |
func |
get_mpu_ranks |
5 | 0 | 0 |
func |
reshape_meg_2d_parallel |
6 | 0 | 0 |
deepspeed.checkpoint.reshape_utils (29 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
partition_data |
3 | 0 | 0 |
func |
merge_state |
4 | 0 | 0 |
func |
sort_zero_files |
3 | 0 | 0 |
func |
merge_state_dict |
4 | 0 | 0 |
func |
get_files_with_prefix |
3 | 0 | 0 |
func |
validate_files |
2 | 0 | 0 |
func |
get_zero_files |
2 | 0 | 0 |
func |
basic_folder_validation |
2 | 0 | 0 |
func |
get_files |
2 | 0 | 0 |
func |
merge_state_list |
4 | 0 | 0 |
deepspeed.checkpoint.universal_checkpoint (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
load_hp_checkpoint_state |
5 | 0 | 0 |
func |
enable_universal_checkpoint |
2 | 0 | 0 |
deepspeed.checkpoint.utils (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_zero_ckpt_name_for_rank |
4 | 0 | 0 |
func |
clone_tensors_for_torch_save |
3 | 0 | 0 |
func |
get_layer_ckpt_name_for_rank |
4 | 0 | 0 |
func |
get_model_ckpt_name_for_rank |
3 | 0 | 0 |
deepspeed.checkpoint.zero_checkpoint (41 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZeROCheckpoint.init |
2 | 0 | 0 |
meth |
ZeROCheckpoint.get_src_world_size |
1 | 0 | 0 |
meth |
ZeROCheckpoint.get_src_tp_degree |
1 | 0 | 0 |
meth |
ZeROCheckpoint.get_src_pp_degree |
1 | 0 | 0 |
meth |
ZeROCheckpoint.get_src_dp_degree |
1 | 0 | 0 |
meth |
ZeROCheckpoint.get_file_indices_for_rank |
4 | 0 | 0 |
meth |
ZeROCheckpoint.get_files_for_rank |
4 | 0 | 0 |
meth |
ZeROCheckpoint.get_state_for_rank |
6 | 0 | 0 |
meth |
ZeROCheckpoint.print_3d_index_map |
2 | 0 | 0 |
meth |
ZeROCheckpoint.print_3d_file_map |
2 | 0 | 0 |
meth |
ZeROCheckpoint.reshape |
2 | 1 | 0 |
meth |
ZeROCheckpoint._strip_tensor_paddings |
2 | 0 | 0 |
meth |
ZeROCheckpoint._clear_group_paddings |
2 | 0 | 0 |
meth |
ZeROCheckpoint._get_optimizer_state |
3 | 0 | 0 |
meth |
ZeROCheckpoint._get_param_group_states |
2 | 0 | 0 |
meth |
ZeROCheckpoint._update_partition_count |
2 | 0 | 0 |
attr |
ZeROCheckpoint.dir |
1 | 0 | 0 |
attr |
ZeROCheckpoint.file_list |
1 | 0 | 0 |
attr |
ZeROCheckpoint.num_files |
1 | 0 | 0 |
attr |
ZeROCheckpoint.src_3d |
1 | 0 | 0 |
attr |
ZeROCheckpoint.target_3d |
1 | 0 | 0 |
deepspeed.comm.backend (13 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Backend.init |
4 | 0 | 0 |
meth |
Backend.is_initialized |
1 | 0 | 0 |
meth |
Backend.new_group |
1 | 0 | 0 |
meth |
Backend.init_process_group |
1 | 0 | 0 |
attr |
Backend.name |
1 | 0 | 0 |
attr |
Backend.world_group |
1 | 0 | 0 |
attr |
Backend.world_size |
1 | 0 | 0 |
attr |
Backend.world_rank |
1 | 0 | 0 |
attr |
Backend.process_groups |
1 | 0 | 0 |
attr |
Backend.initialized |
1 | 0 | 0 |
deepspeed.comm.ccl (106 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CCLHandler.init |
2 | 0 | 0 |
meth |
CCLHandler.wait |
1 | 0 | 0 |
attr |
CCLHandler.ccl_comm_op |
1 | 0 | 0 |
func |
build_ccl_op |
1 | 0 | 0 |
meth |
CCLBackend.init |
7 | 0 | 0 |
meth |
CCLBackend.is_initialized |
1 | 0 | 0 |
meth |
CCLBackend.run_collective |
3 | 0 | 0 |
meth |
CCLBackend.all_reduce |
5 | 0 | 0 |
meth |
CCLBackend.inference_all_reduce |
4 | 0 | 0 |
meth |
CCLBackend.broadcast |
5 | 0 | 0 |
meth |
CCLBackend.all_gather |
5 | 0 | 0 |
meth |
CCLBackend.reduce_scatter_tensor |
6 | 0 | 0 |
meth |
CCLBackend.all_gather_into_tensor |
5 | 0 | 0 |
meth |
CCLBackend.all_to_all_single |
7 | 0 | 0 |
meth |
CCLBackend.send |
5 | 0 | 0 |
meth |
CCLBackend.recv |
5 | 0 | 0 |
meth |
CCLBackend.gather |
6 | 0 | 0 |
meth |
CCLBackend.scatter |
6 | 0 | 0 |
meth |
CCLBackend.barrier |
3 | 0 | 0 |
meth |
CCLBackend.monitored_barrier |
4 | 0 | 0 |
meth |
CCLBackend.reduce_scatter |
6 | 0 | 0 |
meth |
CCLBackend.reduce |
6 | 0 | 0 |
meth |
CCLBackend.new_group |
2 | 0 | 0 |
meth |
CCLBackend._new_group |
3 | 0 | 0 |
meth |
CCLBackend.get_all_ranks_from_group |
2 | 0 | 0 |
attr |
CCLBackend.ccl_comm_op |
1 | 0 | 0 |
attr |
CCLBackend.name |
1 | 0 | 0 |
attr |
CCLBackend.initialized |
1 | 0 | 0 |
attr |
CCLBackend.groups |
1 | 0 | 0 |
attr |
CCLBackend.available_coll |
1 | 0 | 0 |
deepspeed.comm.comm (259 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
in_aml |
1 | 0 | 0 |
func |
scatter |
9 | 0 | 0 |
func |
send |
8 | 0 | 0 |
func |
new_group |
2 | 0 | 0 |
func |
enable_symm_mem_for_group |
2 | 1 | 0 |
func |
monitored_barrier |
7 | 0 | 0 |
func |
all_gather_into_tensor |
8 | 0 | 0 |
func |
inference_all_reduce |
8 | 0 | 0 |
func |
get_local_rank |
1 | 0 | 0 |
func |
irecv |
8 | 0 | 0 |
func |
get_world_size |
2 | 1 | 0 |
meth |
ProcessGroup.init |
3 | 0 | 0 |
attr |
ProcessGroup.ranks |
1 | 0 | 0 |
attr |
ProcessGroup.comm_id |
1 | 0 | 0 |
attr |
ProcessGroup.size |
1 | 0 | 0 |
func |
all_reduce_coalesced |
8 | 0 | 0 |
func |
has_coalescing_manager |
1 | 0 | 0 |
func |
get_all_ranks_from_group |
2 | 0 | 0 |
func |
reset_log |
1 | 0 | 0 |
func |
barrier |
7 | 0 | 0 |
func |
in_aws_sm |
1 | 0 | 0 |
func |
all_to_all_single |
10 | 0 | 0 |
func |
broadcast_object_list |
5 | 0 | 0 |
func |
destroy_process_group |
2 | 0 | 0 |
func |
reduce_scatter_tensor |
9 | 0 | 0 |
func |
set_backend |
1 | 0 | 0 |
func |
get_global_rank |
3 | 0 | 0 |
func |
get_logged_comm_ops |
1 | 0 | 0 |
func |
log_summary |
3 | 0 | 0 |
func |
all_gather_coalesced |
5 | 0 | 0 |
func |
all_gather |
8 | 0 | 0 |
func |
all_to_all |
5 | 0 | 0 |
func |
configure |
7 | 0 | 0 |
func |
reduce_scatter_fn |
8 | 0 | 0 |
func |
isend |
8 | 0 | 0 |
func |
all_reduce |
8 | 0 | 0 |
func |
get_world_group |
1 | 0 | 0 |
func |
mpi_discovery |
3 | 0 | 0 |
func |
init_deepspeed_backend |
4 | 0 | 0 |
func |
broadcast |
8 | 0 | 0 |
func |
gather |
9 | 0 | 0 |
func |
is_initialized |
1 | 0 | 0 |
func |
has_all_gather_into_tensor |
1 | 0 | 0 |
func |
has_all_reduce_coalesced |
1 | 0 | 0 |
func |
reduce |
9 | 0 | 0 |
func |
get_rank |
2 | 0 | 0 |
func |
recv |
8 | 0 | 0 |
func |
in_dlts |
1 | 0 | 0 |
func |
has_comm_data |
1 | 0 | 0 |
func |
patch_aml_env_for_torch_nccl_backend |
3 | 0 | 0 |
attr |
timers |
1 | 0 | 0 |
func |
allgather_fn |
6 | 0 | 0 |
func |
reduce_scatter |
9 | 0 | 0 |
func |
get_comm_operation_count |
1 | 0 | 0 |
func |
timed_op |
2 | 0 | 0 |
func |
all_gather_object |
7 | 0 | 0 |
func |
patch_aws_sm_env_for_torch_nccl_backend |
2 | 0 | 0 |
func |
has_reduce_scatter_tensor |
1 | 0 | 0 |
func |
init_distributed |
11 | 0 | 0 |
attr |
comms_logger |
1 | 0 | 0 |
func |
initialize_mesh_device |
3 | 0 | 0 |
deepspeed.comm.config (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedCommsConfig.init |
2 | 0 | 0 |
attr |
DeepSpeedCommsConfig.comms_logger_enabled |
1 | 0 | 0 |
attr |
DeepSpeedCommsConfig.comms_logger |
1 | 0 | 0 |
deepspeed.comm.torch (188 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TorchBackend.init |
7 | 0 | 0 |
meth |
TorchBackend.get_all_gather_function |
1 | 0 | 0 |
meth |
TorchBackend.get_reduce_scatter_function |
1 | 0 | 0 |
meth |
TorchBackend.has_all_gather_into_tensor |
1 | 0 | 0 |
meth |
TorchBackend.has_reduce_scatter_tensor |
1 | 0 | 0 |
meth |
TorchBackend.init_process_group |
6 | 0 | 0 |
meth |
TorchBackend.all_reduce |
5 | 0 | 0 |
meth |
TorchBackend.inference_all_reduce |
4 | 0 | 0 |
meth |
TorchBackend.all_reduce_coalesced |
5 | 0 | 0 |
meth |
TorchBackend.reduce |
6 | 0 | 0 |
meth |
TorchBackend.reduce_scatter |
6 | 0 | 0 |
meth |
TorchBackend.broadcast |
5 | 0 | 0 |
meth |
TorchBackend.broadcast_object_list |
5 | 0 | 0 |
meth |
TorchBackend.all_gather |
5 | 0 | 0 |
meth |
TorchBackend.all_gather_into_tensor |
5 | 0 | 0 |
meth |
TorchBackend.all_gather_base |
5 | 0 | 0 |
meth |
TorchBackend.all_gather_coalesced |
5 | 0 | 0 |
meth |
TorchBackend.all_gather_object |
4 | 0 | 0 |
meth |
TorchBackend.reduce_scatter_tensor |
6 | 0 | 0 |
meth |
TorchBackend.all_to_all_single |
7 | 0 | 0 |
meth |
TorchBackend.all_to_all |
5 | 0 | 0 |
meth |
TorchBackend.send |
5 | 0 | 0 |
meth |
TorchBackend.recv |
5 | 0 | 0 |
meth |
TorchBackend.isend |
5 | 0 | 0 |
meth |
TorchBackend.irecv |
5 | 0 | 0 |
meth |
TorchBackend.gather |
6 | 0 | 0 |
meth |
TorchBackend.scatter |
6 | 0 | 0 |
meth |
TorchBackend.barrier |
4 | 0 | 0 |
meth |
TorchBackend.monitored_barrier |
4 | 0 | 0 |
meth |
TorchBackend.get_rank |
2 | 0 | 0 |
meth |
TorchBackend.get_world_size |
2 | 0 | 0 |
meth |
TorchBackend.is_initialized |
1 | 0 | 0 |
meth |
TorchBackend.get_backend |
2 | 0 | 0 |
meth |
TorchBackend.new_group |
2 | 0 | 0 |
meth |
TorchBackend.get_global_rank |
3 | 0 | 0 |
meth |
TorchBackend.get_world_group |
1 | 0 | 0 |
meth |
TorchBackend.destroy_process_group |
2 | 0 | 0 |
meth |
TorchBackend._reduce_op |
2 | 0 | 0 |
meth |
TorchBackend.init_device_mesh |
3 | 0 | 0 |
meth |
TorchBackend.enable_symm_mem_for_group |
2 | 0 | 0 |
attr |
TorchBackend.shm_comm_op |
1 | 0 | 0 |
attr |
TorchBackend.has_all_reduce_coalesced |
1 | 0 | 0 |
attr |
TorchBackend.has_coalescing_manager |
1 | 0 | 0 |
attr |
TorchBackend.all_gather_function |
1 | 0 | 0 |
attr |
TorchBackend.reduce_scatter_function |
1 | 0 | 0 |
attr |
TorchBackend.initialized |
1 | 0 | 0 |
attr |
TorchBackend.name |
1 | 0 | 0 |
attr |
TorchBackend.single_gpu_mode |
1 | 0 | 0 |
func |
backward_comm_off |
2 | 0 | 0 |
func |
has_all_reduce_coalesced |
1 | 0 | 0 |
func |
all_gather_comm_off |
2 | 0 | 0 |
func |
reduce_scatter_comm_off |
2 | 0 | 0 |
func |
build_shm_op |
1 | 0 | 0 |
meth |
Noop.wait |
1 | 0 | 0 |
func |
reduce_comm_off |
2 | 0 | 0 |
func |
all_reduce_comm_off |
2 | 0 | 0 |
func |
get_coalescing_manager |
5 | 0 | 0 |
func |
disable_compiler_collective |
2 | 0 | 0 |
func |
broadcast_comm_off |
2 | 0 | 0 |
func |
has_coalescing_manager |
1 | 0 | 0 |
deepspeed.comm.utils (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_msg_size_from_args |
4 | 0 | 0 |
func |
get_world_rank_from_launcher |
1 | 0 | 0 |
func |
get_default_args |
2 | 0 | 0 |
func |
get_world_size_from_launcher |
1 | 0 | 0 |
func |
get_tensor_kwarg |
3 | 0 | 0 |
func |
get_debug_log_name |
3 | 0 | 0 |
func |
get_local_rank_from_launcher |
1 | 0 | 0 |
func |
get_tensor_position |
2 | 0 | 0 |
deepspeed.compat (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_annotations |
2 | 2 | 1 |
deepspeed.compile.backend (29 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
set_example_values_to_symints |
3 | 0 | 0 |
attr |
frames_needing_bwd |
1 | 0 | 0 |
func |
run_opt_passes |
11 | 7 | 0 |
func |
init_schedule |
2 | 0 | 0 |
func |
launch_compile_passes |
2 | 1 | 0 |
func |
evaluate_symint_from_shape_env |
2 | 0 | 0 |
meth |
GraphOrder.init |
1 | 0 | 0 |
meth |
GraphOrder.add_graph |
4 | 3 | 0 |
meth |
GraphOrder.clear |
1 | 0 | 0 |
attr |
GraphOrder.frames |
1 | 0 | 0 |
func |
set_time_and_tensor_size |
6 | 1 | 0 |
func |
register_compile_pass |
3 | 1 | 0 |
attr |
graph_order_with_frame_id |
1 | 0 | 0 |
func |
make_backend |
4 | 0 | 0 |
deepspeed.compile.fx (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
add_postprocess |
8 | 6 | 0 |
func |
move_primals_to_head |
2 | 1 | 0 |
func |
add_args_process |
7 | 5 | 0 |
func |
add_free_activations |
4 | 3 | 0 |
func |
get_output_node |
2 | 1 | 0 |
deepspeed.compile.graph_param (4 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSGraphParam.post_init |
1 | 0 | 0 |
meth |
DSGraphParamManager.init |
4 | 3 | 1 |
meth |
DSGraphParamManager.get_bwd_mapping |
2 | 1 | 0 |
meth |
DSGraphParamManager.get_grad_name |
2 | 1 | 0 |
deepspeed.compile.inductor (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
patch_compiler |
7 | 2 | 0 |
func |
wrap_partition_fn |
4 | 0 | 0 |
func |
patch_create_aot_dispatcher_function |
8 | 2 | 0 |
func |
register_custom_ops |
1 | 0 | 0 |
deepspeed.compile.init_z1 (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
init_z1 |
7 | 0 | 0 |
deepspeed.compile.init_z3 (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
init_z3 |
6 | 0 | 0 |
deepspeed.compile.input_storage (1 missing, 6 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InputStorage.init |
3 | 2 | 0 |
meth |
InputStorage._store_value |
2 | 2 | 2 |
meth |
InputStorage._materialize_value |
2 | 2 | 2 |
meth |
InputStorage.put |
2 | 2 | 1 |
meth |
InputStorage.get |
1 | 1 | 1 |
deepspeed.compile.list_schedule (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
init_schedule |
2 | 1 | 0 |
func |
get_original_args_num |
2 | 1 | 0 |
func |
init_schedule_with_placeholders |
2 | 1 | 0 |
func |
make_graph_from_schedule |
2 | 1 | 0 |
func |
get_new_runnable_nodes_with |
4 | 3 | 0 |
func |
schedule_without_allgather |
4 | 3 | 0 |
func |
choose_next_node |
4 | 3 | 0 |
func |
get_runnable_nodes |
3 | 2 | 0 |
func |
try_schedule_with_new_allgather |
5 | 4 | 0 |
func |
flat_nodes_in_args |
2 | 1 | 0 |
func |
get_node_requirements |
3 | 2 | 0 |
func |
filter_args |
2 | 1 | 0 |
deepspeed.compile.partitioner (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_wrapped_partitioner |
4 | 2 | 0 |
deepspeed.compile.passes (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
run_opt_passes |
12 | 0 | 0 |
deepspeed.compile.passes.offload_adam_states (44 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
update_max_memory |
2 | 0 | 0 |
func |
make_offload_task |
2 | 0 | 0 |
func |
move_opt_states |
9 | 7 | 0 |
func |
make_reload_task |
2 | 0 | 0 |
func |
make_offload_sync |
2 | 0 | 0 |
func |
move_back_hp_param |
4 | 0 | 0 |
func |
sync_reload_states |
2 | 0 | 0 |
func |
move_hp_param |
4 | 0 | 0 |
func |
add_record_max_mem_nodes |
2 | 1 | 0 |
func |
sync_offload_states |
2 | 0 | 0 |
func |
move_back_key |
4 | 0 | 0 |
func |
offload_adam_states_sync |
1 | 0 | 0 |
func |
reload_adam_states_sync |
1 | 0 | 0 |
func |
empty_cache |
1 | 0 | 0 |
func |
lazy_init |
1 | 0 | 0 |
func |
offload_adam_states_for_init |
9 | 7 | 0 |
func |
move_opt_states_sync |
9 | 7 | 0 |
func |
move_key |
4 | 0 | 0 |
func |
print_r0 |
2 | 0 | 0 |
func |
init_offload_opt_states |
3 | 0 | 0 |
deepspeed.compile.passes.offload_parameters (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
add_offload_parameter |
5 | 4 | 0 |
func |
get_ds_id |
2 | 1 | 0 |
func |
offload_parameter_fwd |
9 | 7 | 0 |
func |
add_reload_parameter |
5 | 4 | 0 |
deepspeed.compile.passes.prefetch (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
schedule_prefetch |
9 | 7 | 0 |
func |
get_ds_id |
2 | 1 | 0 |
func |
print_rank_0 |
2 | 0 | 0 |
deepspeed.compile.passes.selective_gather (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
selective_gather |
9 | 7 | 0 |
deepspeed.compile.passes.zero1_compile (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
add_z1_reduce_fw |
6 | 3 | 0 |
func |
add_z1_reduce |
9 | 6 | 0 |
func |
add_z1_reduce_bw |
4 | 3 | 0 |
func |
add_z2_reduce |
9 | 6 | 0 |
deepspeed.compile.passes.zero3_compile (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
add_gather_and_reduce |
6 | 5 | 0 |
func |
add_z3_gather_release_bw |
8 | 4 | 0 |
func |
add_reduce |
6 | 5 | 0 |
func |
add_gather_and_release |
5 | 4 | 0 |
func |
add_z3_gather_release |
9 | 6 | 0 |
func |
add_release |
7 | 6 | 0 |
func |
add_z3_gather_release_fw |
8 | 4 | 0 |
func |
add_allgather |
6 | 5 | 0 |
deepspeed.compile.patch_compiled_func (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
patch_compiled_func |
1 | 0 | 0 |
func |
unpatch_compiled_func |
1 | 0 | 0 |
func |
get_backward_inputs |
1 | 0 | 0 |
attr |
base_meta |
1 | 0 | 0 |
meth |
FunctionMeta.new |
4 | 0 | 0 |
deepspeed.compile.patch_fake_tensor (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
wrap_if_ds_param |
2 | 0 | 0 |
func |
patch_fake_tensor |
1 | 0 | 0 |
deepspeed.compile.profilers.comm_profile (25 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
timed_all_gather |
9 | 0 | 0 |
attr |
local_rank |
1 | 0 | 0 |
attr |
predictor |
1 | 0 | 0 |
attr |
predicted_time |
1 | 0 | 0 |
func |
run_all_gather |
7 | 0 | 0 |
func |
create_predictor |
1 | 0 | 0 |
func |
sync_all |
1 | 0 | 0 |
func |
get_bw |
4 | 0 | 0 |
deepspeed.compile.profilers.graph_profile (22 missing, 4 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MemoryProfilingInterpreter.init |
3 | 1 | 0 |
meth |
MemoryProfilingInterpreter.run |
2 | 1 | 1 |
meth |
MemoryProfilingInterpreter.run_node |
2 | 2 | 1 |
meth |
MemoryProfilingInterpreter.dump |
2 | 0 | 0 |
attr |
MemoryProfilingInterpreter.nz3 |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.device |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.mem_record |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.last_alloc |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.node_counter |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.node_num |
1 | 0 | 0 |
attr |
MemoryProfilingInterpreter.debug_log |
1 | 0 | 0 |
meth |
ProfilingInterpreter.init |
5 | 3 | 0 |
meth |
ProfilingInterpreter.run |
2 | 1 | 1 |
meth |
ProfilingInterpreter.run_node |
2 | 2 | 1 |
attr |
ProfilingInterpreter.nz3 |
1 | 0 | 0 |
attr |
ProfilingInterpreter.iteration |
1 | 0 | 0 |
attr |
ProfilingInterpreter.warmup |
1 | 0 | 0 |
attr |
ProfilingInterpreter.device |
1 | 0 | 0 |
attr |
ProfilingInterpreter.distributed |
1 | 0 | 0 |
attr |
ProfilingInterpreter.debug_log |
1 | 0 | 0 |
attr |
ProfilingInterpreter.mem_usage_out_of_torch |
1 | 0 | 0 |
deepspeed.compile.util (40 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
add_pre_backward_hook |
2 | 0 | 0 |
func |
count_inflight_values |
3 | 2 | 0 |
func |
tensor_meta_size |
2 | 1 | 0 |
func |
get_last_uses |
2 | 1 | 0 |
func |
get_no_copy_ops |
1 | 0 | 0 |
meth |
NodeValueOffloadHelper.init |
2 | 0 | 0 |
meth |
NodeValueOffloadHelper._to_cpu |
2 | 0 | 0 |
meth |
NodeValueOffloadHelper._from_cpu |
2 | 0 | 0 |
meth |
NodeValueOffloadHelper.save |
4 | 3 | 0 |
attr |
NodeValueOffloadHelper.device |
1 | 0 | 0 |
func |
get_deepcompile_handle |
1 | 0 | 0 |
func |
deepcompile_backward_prologue |
2 | 0 | 0 |
func |
is_backend_inductor |
2 | 0 | 0 |
func |
log_rank0 |
3 | 2 | 0 |
func |
get_activation_node_names |
4 | 3 | 0 |
func |
add_mem_profile_nodes |
3 | 2 | 0 |
meth |
TensorOffloadHelper.init |
1 | 0 | 0 |
meth |
TensorOffloadHelper.offload |
2 | 0 | 0 |
meth |
TensorOffloadHelper.reload |
2 | 0 | 0 |
attr |
TensorOffloadHelper.devices |
1 | 0 | 0 |
attr |
TensorOffloadHelper.base_tensors |
1 | 0 | 0 |
attr |
TensorOffloadHelper.views |
1 | 0 | 0 |
attr |
TensorOffloadHelper.arg_list |
1 | 0 | 0 |
attr |
TensorOffloadHelper.offloaded |
1 | 0 | 0 |
attr |
TensorOffloadHelper.non_tensor |
1 | 0 | 0 |
func |
get_real_uses |
2 | 1 | 0 |
func |
materialize_fake |
3 | 0 | 0 |
func |
get_index_by_graph_id |
3 | 0 | 0 |
deepspeed.compression.basic_layer (156 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
gather_from_model_parallel_region |
2 | 0 | 0 |
meth |
RowParallelLinear_Compress.init |
7 | 0 | 0 |
meth |
RowParallelLinear_Compress.forward |
2 | 0 | 0 |
attr |
RowParallelLinear_Compress.input_size |
1 | 0 | 0 |
attr |
RowParallelLinear_Compress.output_size |
1 | 0 | 0 |
attr |
RowParallelLinear_Compress.input_is_parallel |
1 | 0 | 0 |
attr |
RowParallelLinear_Compress.skip_bias_add |
1 | 0 | 0 |
attr |
RowParallelLinear_Compress.input_size_per_partition |
1 | 0 | 0 |
func |
scatter_to_model_parallel_region |
2 | 0 | 0 |
meth |
Embedding_Compress.init |
2 | 0 | 0 |
meth |
Embedding_Compress.extra_repr |
1 | 0 | 0 |
meth |
Embedding_Compress.enable_weight_quantization |
7 | 0 | 0 |
meth |
Embedding_Compress.fix_weight_quantization |
1 | 0 | 0 |
meth |
Embedding_Compress.forward |
2 | 0 | 0 |
attr |
Embedding_Compress.weight_quantization_enabled_in_forward |
1 | 0 | 0 |
attr |
Embedding_Compress.weight_quantization_enabled |
1 | 0 | 0 |
meth |
LinearLayer_Compress.init |
3 | 0 | 0 |
meth |
LinearLayer_Compress.extra_repr |
1 | 0 | 0 |
meth |
LinearLayer_Compress.enable_sparse_pruning |
3 | 0 | 0 |
meth |
LinearLayer_Compress.enable_row_pruning |
3 | 0 | 0 |
meth |
LinearLayer_Compress.enable_head_pruning |
4 | 0 | 0 |
meth |
LinearLayer_Compress.fix_sparse_pruning_helper |
1 | 0 | 0 |
meth |
LinearLayer_Compress.fix_row_col_pruning_helper |
3 | 0 | 0 |
meth |
LinearLayer_Compress.fix_head_pruning_helper |
4 | 0 | 0 |
meth |
LinearLayer_Compress.get_mask |
2 | 0 | 0 |
meth |
LinearLayer_Compress.enable_weight_quantization |
7 | 0 | 0 |
meth |
LinearLayer_Compress.fix_weight_quantization |
1 | 0 | 0 |
meth |
LinearLayer_Compress.enable_activation_quantization |
4 | 0 | 0 |
meth |
LinearLayer_Compress.head_pruning_reshape |
3 | 0 | 0 |
meth |
LinearLayer_Compress.forward |
3 | 0 | 0 |
attr |
LinearLayer_Compress.sparse_pruning_method |
1 | 0 | 0 |
attr |
LinearLayer_Compress.row_pruning_method |
1 | 0 | 0 |
attr |
LinearLayer_Compress.head_pruning_method |
1 | 0 | 0 |
attr |
LinearLayer_Compress.activation_quantization_method |
1 | 0 | 0 |
attr |
LinearLayer_Compress.weight_quantization_enabled_in_forward |
1 | 0 | 0 |
attr |
LinearLayer_Compress.weight_quantization_enabled |
1 | 0 | 0 |
attr |
LinearLayer_Compress.sparse_pruning_enabled |
1 | 0 | 0 |
attr |
LinearLayer_Compress.row_pruning_enabled |
1 | 0 | 0 |
attr |
LinearLayer_Compress.head_pruning_enabled |
1 | 0 | 0 |
attr |
LinearLayer_Compress.activation_quantization_enabled |
1 | 0 | 0 |
func |
copy_to_model_parallel_region |
2 | 0 | 0 |
meth |
ColumnParallelLinear_Compress.init |
7 | 0 | 0 |
meth |
ColumnParallelLinear_Compress.forward |
2 | 0 | 0 |
attr |
ColumnParallelLinear_Compress.input_size |
1 | 0 | 0 |
attr |
ColumnParallelLinear_Compress.output_size |
1 | 0 | 0 |
attr |
ColumnParallelLinear_Compress.gather_output |
1 | 0 | 0 |
attr |
ColumnParallelLinear_Compress.skip_bias_add |
1 | 0 | 0 |
attr |
ColumnParallelLinear_Compress.output_size_per_partition |
1 | 0 | 0 |
meth |
QuantAct.init |
3 | 0 | 0 |
meth |
QuantAct.forward |
4 | 0 | 0 |
attr |
QuantAct.act_range_momentum |
1 | 0 | 0 |
attr |
QuantAct.quant_mode |
1 | 0 | 0 |
attr |
QuantAct.act_function |
1 | 0 | 0 |
meth |
Conv2dLayer_Compress.init |
2 | 0 | 0 |
meth |
Conv2dLayer_Compress.repr |
1 | 0 | 0 |
meth |
Conv2dLayer_Compress.enable_sparse_pruning |
3 | 0 | 0 |
meth |
Conv2dLayer_Compress.enable_channel_pruning |
3 | 0 | 0 |
meth |
Conv2dLayer_Compress.fix_sparse_pruning_helper |
1 | 0 | 0 |
meth |
Conv2dLayer_Compress.fix_channel_pruning_helper |
3 | 0 | 0 |
meth |
Conv2dLayer_Compress.get_mask |
2 | 0 | 0 |
meth |
Conv2dLayer_Compress.fix_weight_quantization |
1 | 0 | 0 |
meth |
Conv2dLayer_Compress.enable_weight_quantization |
7 | 0 | 0 |
meth |
Conv2dLayer_Compress.enable_activation_quantization |
4 | 0 | 0 |
meth |
Conv2dLayer_Compress.forward |
2 | 0 | 0 |
attr |
Conv2dLayer_Compress.sparse_pruning_method |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.channel_pruning_method |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.activation_quantization_method |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.weight_quantization_enabled_in_forward |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.sparse_pruning_enabled |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.channel_pruning_enabled |
1 | 0 | 0 |
attr |
Conv2dLayer_Compress.activation_quantization_enabled |
1 | 0 | 0 |
func |
reduce_from_model_parallel_region |
2 | 0 | 0 |
meth |
BNLayer_Compress.fix_channel_pruning_helper |
3 | 0 | 0 |
func |
split_tensor_along_last_dim |
4 | 0 | 0 |
deepspeed.compression.compress (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_compress_methods |
4 | 0 | 0 |
func |
init_compression |
5 | 0 | 0 |
func |
redundancy_clean |
4 | 0 | 0 |
func |
student_initialization |
4 | 0 | 0 |
func |
check_deepspeed_config |
2 | 0 | 0 |
func |
get_module_name |
7 | 0 | 0 |
deepspeed.compression.config (46 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_activation_quantization_different_groups |
2 | 0 | 0 |
func |
get_weight_quantization_different_groups |
2 | 0 | 0 |
func |
get_weight_quantization_shared_parameters |
2 | 0 | 0 |
func |
get_row_pruning_shared_parameters |
2 | 0 | 0 |
func |
get_channel_pruning_shared_parameters |
2 | 0 | 0 |
func |
get_row_pruning |
2 | 0 | 0 |
func |
get_layer_reduction_params |
2 | 0 | 0 |
func |
get_head_pruning_shared_parameters |
2 | 0 | 0 |
func |
get_quantize_enabled |
2 | 0 | 0 |
func |
get_weight_quantization |
2 | 0 | 0 |
func |
get_sparse_pruning_different_groups |
2 | 0 | 0 |
func |
get_activation_quantization |
2 | 0 | 0 |
func |
get_sparse_pruning_shared_parameters |
2 | 0 | 0 |
func |
get_compression_config |
2 | 0 | 0 |
func |
get_channel_pruning_different_groups |
2 | 0 | 0 |
func |
get_activation_quantization_shared_parameters |
2 | 0 | 0 |
func |
get_layer_reduction_enabled |
2 | 0 | 0 |
func |
get_head_pruning |
2 | 0 | 0 |
func |
get_channel_pruning |
2 | 0 | 0 |
func |
get_head_pruning_different_groups |
2 | 0 | 0 |
func |
get_sparse_pruning |
2 | 0 | 0 |
func |
get_row_pruning_different_groups |
2 | 0 | 0 |
func |
get_layer_reduction |
2 | 0 | 0 |
deepspeed.compression.helper (34 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
recursive_getattr |
3 | 0 | 0 |
func |
compression_preparation |
4 | 0 | 0 |
func |
fix_compression |
6 | 0 | 0 |
func |
convert_conv1d_to_linear |
3 | 0 | 0 |
func |
module_replacement |
5 | 0 | 0 |
func |
recursive_setattr |
4 | 0 | 0 |
func |
generate_pruners |
3 | 0 | 0 |
func |
is_module_compressible |
3 | 0 | 0 |
func |
rewrite_optimizer_step |
2 | 1 | 0 |
func |
register_on_step_begin |
2 | 0 | 0 |
deepspeed.compression.scheduler (18 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
compression_scheduler.init |
3 | 0 | 0 |
meth |
compression_scheduler.make_init |
1 | 0 | 0 |
meth |
compression_scheduler.check_weight_quantization |
1 | 0 | 0 |
meth |
compression_scheduler.check_activation_quantization |
1 | 0 | 0 |
meth |
compression_scheduler.check_sparse_pruning |
1 | 0 | 0 |
meth |
compression_scheduler.check_head_pruning |
1 | 0 | 0 |
meth |
compression_scheduler.check_row_pruning |
1 | 0 | 0 |
meth |
compression_scheduler.check_channel_pruning |
1 | 0 | 0 |
meth |
compression_scheduler.check_all_modules |
1 | 0 | 0 |
meth |
compression_scheduler.step |
2 | 0 | 0 |
attr |
compression_scheduler.model |
1 | 0 | 0 |
attr |
compression_scheduler.compression_config |
1 | 0 | 0 |
attr |
compression_scheduler.training_steps |
1 | 0 | 0 |
attr |
compression_scheduler.weight_quantization_enabled |
1 | 0 | 0 |
attr |
compression_scheduler.verbose |
1 | 0 | 0 |
deepspeed.compression.utils (45 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TernaryQuantizer.forward |
7 | 0 | 0 |
meth |
TernaryQuantizer.backward |
3 | 0 | 0 |
meth |
SymQuantizer.forward |
7 | 0 | 0 |
meth |
SymQuantizer.backward |
3 | 0 | 0 |
meth |
AsymQuantizer.forward |
7 | 0 | 0 |
meth |
AsymQuantizer.backward |
3 | 0 | 0 |
meth |
TopKBinarizer.forward |
5 | 3 | 0 |
meth |
TopKBinarizer.backward |
3 | 0 | 0 |
meth |
BinaryQuantizer.forward |
7 | 0 | 0 |
meth |
BinaryQuantizer.backward |
3 | 0 | 0 |
deepspeed.constants (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
default_pg_timeout |
1 | 0 | 0 |
deepspeed.datastates.config (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedDataStatesConfig.init |
2 | 0 | 0 |
attr |
DeepSpeedDataStatesConfig.enabled |
1 | 0 | 0 |
attr |
DeepSpeedDataStatesConfig.config |
1 | 0 | 0 |
deepspeed.elasticity.config (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ElasticityConfig.init |
2 | 0 | 0 |
meth |
ElasticityConfig.repr |
1 | 0 | 0 |
meth |
ElasticityConfig.repr |
1 | 0 | 0 |
attr |
ElasticityConfig.enabled |
1 | 0 | 0 |
attr |
ElasticityConfig.min_gpus |
1 | 0 | 0 |
attr |
ElasticityConfig.max_gpus |
1 | 0 | 0 |
attr |
ElasticityConfig.model_parallel_size |
1 | 0 | 0 |
attr |
ElasticityConfig.num_gpus_per_node |
1 | 0 | 0 |
attr |
ElasticityConfig.min_time |
1 | 0 | 0 |
attr |
ElasticityConfig.version |
1 | 0 | 0 |
attr |
ElasticityConfig.prefer_larger_batch_size |
1 | 0 | 0 |
attr |
ElasticityConfig.ignore_non_elastic_batch_info |
1 | 0 | 0 |
attr |
ElasticityConfig.max_acceptable_batch_size |
1 | 0 | 0 |
attr |
ElasticityConfig.micro_batches |
1 | 0 | 0 |
deepspeed.elasticity.elastic_agent (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
log |
1 | 0 | 0 |
meth |
DSElasticAgent.init |
6 | 4 | 0 |
meth |
DSElasticAgent._set_master_addr_port |
5 | 4 | 0 |
attr |
DSElasticAgent.ds_env |
1 | 0 | 0 |
deepspeed.elasticity.elasticity (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_best_candidates |
6 | 0 | 0 |
func |
compute_elastic_config |
5 | 2 | 0 |
func |
ensure_immutable_elastic_config |
2 | 1 | 0 |
func |
elasticity_enabled |
2 | 1 | 0 |
func |
get_valid_gpus |
5 | 0 | 0 |
func |
get_candidate_batch_sizes |
3 | 0 | 0 |
deepspeed.elasticity.utils (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
is_torch_elastic_compatible |
1 | 0 | 0 |
deepspeed.env_report (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
cli_main |
1 | 0 | 0 |
func |
op_report |
2 | 0 | 0 |
func |
installed_cann_version |
1 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
func |
debug_report |
1 | 0 | 0 |
func |
nvcc_version |
1 | 0 | 0 |
func |
main |
3 | 0 | 0 |
func |
get_shm_size |
1 | 0 | 0 |
func |
human_readable_size |
2 | 0 | 0 |
attr |
color_len |
1 | 0 | 0 |
func |
installed_cann_path |
1 | 0 | 0 |
deepspeed.git_version_info (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
op_compatible |
1 | 0 | 0 |
attr |
version |
1 | 0 | 0 |
attr |
compatible_ops |
1 | 0 | 0 |
attr |
installed_ops |
1 | 0 | 0 |
deepspeed.inference.config (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DtypeEnum.from_str |
2 | 1 | 0 |
meth |
DeepSpeedInferenceConfig.validate_dtype |
3 | 0 | 0 |
meth |
DeepSpeedInferenceConfig.moe_backward_compat |
3 | 0 | 0 |
meth |
DeepSpeedInferenceConfig.has_triton |
3 | 0 | 0 |
deepspeed.inference.engine (77 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceEngine.init |
3 | 0 | 0 |
meth |
InferenceEngine.destroy |
1 | 0 | 0 |
meth |
InferenceEngine.profile_model_time |
2 | 0 | 0 |
meth |
InferenceEngine._get_model_config_generate |
2 | 0 | 0 |
meth |
InferenceEngine.remove_mask_prepare_for_bloom |
1 | 0 | 0 |
meth |
InferenceEngine.build_alibi_tensor |
1 | 0 | 0 |
meth |
InferenceEngine.build_attn_bias |
1 | 0 | 0 |
meth |
InferenceEngine._pre_forward_hook |
4 | 0 | 0 |
meth |
InferenceEngine._post_forward_hook |
4 | 0 | 0 |
meth |
InferenceEngine._create_model_parallel_group |
2 | 0 | 0 |
meth |
InferenceEngine._create_ep_parallel_group |
2 | 0 | 0 |
meth |
InferenceEngine._init_quantization_setting |
2 | 0 | 0 |
meth |
InferenceEngine.load_model_with_checkpoint |
2 | 0 | 0 |
meth |
InferenceEngine._apply_injection_policy |
3 | 0 | 0 |
meth |
InferenceEngine._get_all_ckpt_names |
3 | 0 | 0 |
meth |
InferenceEngine._get_ckpt_name |
4 | 0 | 0 |
meth |
InferenceEngine._load_checkpoint |
4 | 0 | 0 |
meth |
InferenceEngine._choose_module_key |
2 | 0 | 0 |
meth |
InferenceEngine._convert_to_dtype |
2 | 0 | 0 |
meth |
InferenceEngine._create_cuda_graph |
3 | 0 | 0 |
meth |
InferenceEngine._graph_replay |
3 | 0 | 0 |
meth |
InferenceEngine.model_times |
1 | 0 | 0 |
meth |
InferenceEngine._module_match |
2 | 0 | 0 |
meth |
InferenceEngine._local_cuda_graph_used |
2 | 0 | 0 |
meth |
InferenceEngine.forward |
3 | 0 | 0 |
meth |
InferenceEngine._generate |
3 | 0 | 0 |
meth |
InferenceEngine.compile |
3 | 1 | 0 |
attr |
InferenceEngine.expert_mp_group |
1 | 0 | 0 |
attr |
InferenceEngine.module |
1 | 0 | 0 |
attr |
InferenceEngine.injection_dict |
1 | 0 | 0 |
attr |
InferenceEngine.mp_group |
1 | 0 | 0 |
attr |
InferenceEngine.mpu |
1 | 0 | 0 |
attr |
InferenceEngine.quantize_merge_count |
1 | 0 | 0 |
attr |
InferenceEngine.quantization_scales |
1 | 0 | 0 |
attr |
InferenceEngine.ep_group |
1 | 0 | 0 |
attr |
InferenceEngine.cuda_graph_created |
1 | 0 | 0 |
attr |
InferenceEngine.checkpoint_engine |
1 | 0 | 0 |
attr |
InferenceEngine.model_profile_enabled |
1 | 0 | 0 |
attr |
InferenceEngine.local_cuda_graph |
1 | 0 | 0 |
attr |
InferenceEngine.generate |
1 | 0 | 0 |
deepspeed.inference.quantization.layers (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_quantized_weight_wrapper |
4 | 3 | 0 |
attr |
QuantizedLinear.config |
1 | 0 | 0 |
attr |
QuantizedLinear.quantizer |
1 | 0 | 0 |
attr |
QuantizedLinear.bias |
1 | 0 | 0 |
attr |
QuantizedLinear.weight |
1 | 0 | 0 |
attr |
QuantizedEmbedding.config |
1 | 0 | 0 |
attr |
QuantizedEmbedding.weight |
1 | 0 | 0 |
deepspeed.inference.quantization.quantization_context (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
QuantizationContext.init |
3 | 2 | 0 |
deepspeed.inference.quantization.utils (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
Quantizer.config |
1 | 0 | 0 |
attr |
DeQuantizer.config |
1 | 0 | 0 |
attr |
DeQuantizer.dtype |
1 | 0 | 0 |
attr |
device |
1 | 0 | 0 |
func |
get_quantizer_module |
1 | 0 | 0 |
func |
recursive_setattr |
4 | 0 | 0 |
func |
tensor_clamp |
4 | 2 | 0 |
func |
wrap_load_from_state_dict |
2 | 0 | 0 |
func |
wrap_quantized_functional |
2 | 0 | 0 |
func |
get_AsyncPartitionedParameterSwapper |
2 | 1 | 0 |
deepspeed.inference.v2.allocator (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
on_device |
2 | 1 | 0 |
attr |
Allocator.cache |
1 | 0 | 0 |
deepspeed.inference.v2.checkpoint.huggingface_engine (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
engine |
1 | 0 | 0 |
meth |
HuggingFaceCheckpointEngine.init |
4 | 3 | 0 |
meth |
HuggingFaceCheckpointEngine._fetch_checkpoint_files |
1 | 0 | 0 |
attr |
HuggingFaceCheckpointEngine.model_name_or_path |
1 | 0 | 0 |
attr |
HuggingFaceCheckpointEngine.auth_token |
1 | 0 | 0 |
attr |
HuggingFaceCheckpointEngine.model_config |
1 | 0 | 0 |
deepspeed.inference.v2.checkpoint.in_memory_engine (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
InMemoryModelEngine.model |
1 | 0 | 0 |
deepspeed.inference.v2.engine_v2 (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceEngineV2._initialize_tp_group |
1 | 0 | 0 |
meth |
InferenceEngineV2.query |
4 | 3 | 0 |
deepspeed.inference.v2.inference_parameter (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceParameter.new |
5 | 0 | 0 |
meth |
InferenceParameter.to |
3 | 0 | 0 |
meth |
InferenceParameter.initialize |
3 | 2 | 0 |
meth |
InferenceParameter.initialize_raw |
2 | 1 | 0 |
deepspeed.inference.v2.inference_utils (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DtypeEnum.new |
2 | 0 | 0 |
meth |
DtypeEnum.repr |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.bias_activations.bias_activation (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
CUDABiasActivation.kernel |
1 | 0 | 0 |
attr |
CUDABiasActivation.act_fn |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.blas_kernels.blas_linear (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BlasLibLinear.init |
2 | 1 | 0 |
attr |
BlasLibLinear.inf_module |
1 | 0 | 0 |
attr |
BlasLibLinear.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.cuda_layer_norm.cuda_fp_ln_base (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CUDAFPLNBase.init |
4 | 3 | 0 |
attr |
CUDAFPLNBase.inf_module |
1 | 0 | 0 |
attr |
CUDAFPLNBase.epsilon |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.cuda_linear.cuda_linear (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CUDAWf6Af16Linear.init |
1 | 0 | 0 |
meth |
CUDAWf6Af16Linear.call |
9 | 6 | 0 |
meth |
CUDAWf6Af16Linear.get_workspace |
7 | 5 | 0 |
attr |
CUDAWf6Af16Linear.inf_module |
1 | 0 | 0 |
attr |
CUDAWf6Af16Linear.kernel |
1 | 0 | 0 |
attr |
CUDAWf6Af16Linear.split_k_map |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.cuda_rms_norm.rms_norm_base (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CUDARMSNormBase.init |
4 | 3 | 0 |
attr |
CUDARMSNormBase.inf_module |
1 | 0 | 0 |
attr |
CUDARMSNormBase.epsilon |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.core_ops.gated_activations.gated_activation (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
CUDAGatedActivation.inf_module |
1 | 0 | 0 |
attr |
CUDAGatedActivation.act_fn |
1 | 0 | 0 |
attr |
CUDAGatedActivation.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.cutlass_ops.mixed_gemm.mixed_gemm (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MixedGEMM.num_bits |
1 | 0 | 0 |
attr |
MixedGEMM.kernel |
1 | 0 | 0 |
attr |
MixedGEMM.act_fn |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.cutlass_ops.moe_gemm.mixed_moe_gemm (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MixedMoEGEMM.num_bits |
1 | 0 | 0 |
attr |
MixedMoEGEMM.kernel |
1 | 0 | 0 |
attr |
MixedMoEGEMM.act_fn |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.cutlass_ops.moe_gemm.moe_gemm (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MoEGEMM.kernel |
1 | 0 | 0 |
attr |
MoEGEMM.act_fn |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ds_kernel (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSKernelBase.init |
3 | 0 | 0 |
meth |
DSKernelBase.call |
3 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.atom_builder.atom_builder (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
AtomBuilder.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.blocked_flash.blocked_flash (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
BlockedFlashAttn.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.embed.embed (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
RaggedEmbeddingKernel.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.linear_blocked_kv_rotary.blocked_kv_rotary (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
BlockedRotaryEmbeddings.kernel |
1 | 0 | 0 |
attr |
BlockedRotaryEmbeddings.head_size |
1 | 0 | 0 |
attr |
BlockedRotaryEmbeddings.n_q_heads |
1 | 0 | 0 |
attr |
BlockedRotaryEmbeddings.n_kv_heads |
1 | 0 | 0 |
attr |
BlockedRotaryEmbeddings.rotary_dim |
1 | 0 | 0 |
attr |
BlockedRotaryEmbeddings.theta_base |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.linear_blocked_kv_rotary.blocked_trained_kv_rotary (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
BlockedTrainedRotaryEmbeddings.kernel |
1 | 0 | 0 |
attr |
BlockedTrainedRotaryEmbeddings.head_size |
1 | 0 | 0 |
attr |
BlockedTrainedRotaryEmbeddings.n_q_heads |
1 | 0 | 0 |
attr |
BlockedTrainedRotaryEmbeddings.n_kv_heads |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.linear_blocked_kv_rotary.linear_blocked_kv_copy (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
LinearBlockedKVCopy.kernel |
1 | 0 | 0 |
attr |
LinearBlockedKVCopy.head_size |
1 | 0 | 0 |
attr |
LinearBlockedKVCopy.n_q_heads |
1 | 0 | 0 |
attr |
LinearBlockedKVCopy.n_kv_heads |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.logits_gather.logits_gather (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RaggedLogitsGather.init |
3 | 2 | 0 |
attr |
RaggedLogitsGather.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.moe_gather.moe_gather (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MoEGather.kernel |
1 | 0 | 0 |
attr |
MoEGather.normalize_scores |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.moe_scatter.moe_scatter (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MoEScatter.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.kernels.ragged_ops.top_k_gating.top_k_gating (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
RaggedTopKGating.kernel |
1 | 0 | 0 |
deepspeed.inference.v2.model_implementations.common_parameters.qkv_parameters (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
UnfusedQKVParameter.finalize |
1 | 0 | 0 |
deepspeed.inference.v2.model_implementations.exaone4.model (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Exaone4InferenceModel.init |
3 | 0 | 0 |
deepspeed.inference.v2.model_implementations.exaone4.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Exaone4Policy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.falcon.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FalconPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.inference_model_base (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
cached_property |
2 | 0 | 0 |
meth |
DSInferenceModelBase.set_parameters |
5 | 4 | 0 |
meth |
DSInferenceModelBase.set_state_manager |
2 | 1 | 0 |
prop |
DSInferenceModelBase.model_config |
1 | 0 | 0 |
prop |
DSInferenceModelBase.engine_config |
1 | 0 | 0 |
deepspeed.inference.v2.model_implementations.inference_policy_base (6 missing, 2 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PolicyMeta.new |
4 | 0 | 0 |
meth |
InferenceV2Policy.init |
4 | 4 | 1 |
meth |
InferenceV2Policy.build_model |
3 | 3 | 1 |
meth |
ContainerMap.map_param |
3 | 1 | 0 |
deepspeed.inference.v2.model_implementations.inference_transformer_base (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
cached_property |
2 | 0 | 0 |
deepspeed.inference.v2.model_implementations.layer_container_base (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
LayerMetaclass.new |
4 | 0 | 0 |
meth |
LayerMetaclass.call |
3 | 0 | 0 |
func |
make_finalization_callback |
2 | 1 | 0 |
deepspeed.inference.v2.model_implementations.llama_v2.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Llama2Policy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.mistral.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MistralPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.mixtral.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MixtralPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.opt.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OPTPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.parameter_base (25 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
make_param_getter |
3 | 0 | 0 |
func |
make_readonly_setter |
1 | 0 | 0 |
meth |
ParameterMetaclass.new |
4 | 0 | 0 |
meth |
ParameterMetaclass.call |
3 | 0 | 0 |
attr |
ParameterBase.inference_model |
1 | 0 | 0 |
attr |
ParameterBase.completed_components |
1 | 0 | 0 |
attr |
ParameterBase.parent_container |
1 | 0 | 0 |
func |
make_param_setter |
3 | 0 | 0 |
func |
ParamList |
2 | 1 | 0 |
meth |
ParametrizedList.getitem |
2 | 0 | 0 |
meth |
ParametrizedList.setitem |
3 | 0 | 0 |
meth |
ParametrizedList.iter |
1 | 0 | 0 |
attr |
ParametrizedList.set_params |
1 | 0 | 0 |
deepspeed.inference.v2.model_implementations.phi.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PhiPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.phi3.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Phi3Policy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.qwen.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
QwenPolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.qwen_v2.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Qwen2Policy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.model_implementations.qwen_v2_moe.policy (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Qwen2MoePolicy.instantiate_model |
3 | 3 | 1 |
deepspeed.inference.v2.modules.implementations.attention.dense_blocked_attention (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
cached_property |
2 | 0 | 0 |
attr |
DSDenseBlockedAttention.model_dim |
1 | 0 | 0 |
deepspeed.inference.v2.modules.implementations.embedding.ragged_embedding (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSRaggedEmbedding.name |
1 | 0 | 0 |
attr |
DSRaggedEmbedding.embed_offset |
1 | 0 | 0 |
deepspeed.inference.v2.modules.implementations.linear.blas_fp_linear (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BlasFPLinear.name |
1 | 0 | 0 |
deepspeed.inference.v2.modules.implementations.linear.quantized_linear (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
fp_quantize |
7 | 6 | 0 |
meth |
QuantizedWf6Af16Linear.name |
1 | 0 | 0 |
attr |
QuantizedWf6Af16Linear.inf_module |
1 | 0 | 0 |
attr |
QuantizedWf6Af16Linear.preprocess_weight |
1 | 0 | 0 |
attr |
QuantizedWf6Af16Linear.quantizer |
1 | 0 | 0 |
attr |
QuantizedWf6Af16Linear.out_channels |
1 | 0 | 0 |
attr |
QuantizedWf6Af16Linear.in_channels |
1 | 0 | 0 |
deepspeed.inference.v2.modules.implementations.moe.cutlass_multi_gemm (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSMultiGemmMoE.name |
1 | 0 | 0 |
meth |
DSMultiGemmMoE._create_buffers |
1 | 0 | 0 |
attr |
DSMultiGemmMoE.max_tokens |
1 | 0 | 0 |
attr |
DSMultiGemmMoE.n_experts |
1 | 0 | 0 |
attr |
DSMultiGemmMoE.n_top_k |
1 | 0 | 0 |
attr |
DSMultiGemmMoE.intermediate_dim |
1 | 0 | 0 |
deepspeed.inference.v2.modules.implementations.post_norm.cuda_post_ln (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSPostLNCUDAModule.name |
1 | 0 | 0 |
meth |
DSPostLNCUDAModule.supports_config |
2 | 1 | 0 |
meth |
DSPostLNCUDAModule.init |
3 | 2 | 0 |
deepspeed.inference.v2.modules.implementations.pre_norm.cuda_pre_ln (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSPreLNCUDAModule.name |
1 | 0 | 0 |
meth |
DSPreLNCUDAModule.supports_config |
2 | 1 | 0 |
meth |
DSPreLNCUDAModule.init |
3 | 2 | 0 |
deepspeed.inference.v2.modules.implementations.pre_norm.cuda_pre_rms (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSPreRMSCUDAModule.name |
1 | 0 | 0 |
meth |
DSPreRMSCUDAModule.supports_config |
2 | 1 | 0 |
meth |
DSPreRMSCUDAModule.init |
3 | 2 | 0 |
deepspeed.inference.v2.modules.implementations.unembed.ragged_unembed (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSRaggedUnembed.name |
1 | 0 | 0 |
meth |
DSRaggedUnembed.supports_config |
2 | 1 | 0 |
deepspeed.inference.v2.modules.interfaces.pre_norm_base (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSPreNormBase.init |
3 | 2 | 0 |
deepspeed.inference.v2.ragged.manager_configs (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSStateManagerConfig.max_ragged_sequence_count_validator |
1 | 0 | 0 |
deepspeed.inference.v2.ragged.ragged_wrapper (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RaggedBatchWrapper.insert_sequence |
4 | 3 | 0 |
deepspeed.inference.v2.ragged.sequence_descriptor (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PlaceholderSequenceDescriptor.init |
4 | 1 | 0 |
deepspeed.io.base_file_writer (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BaseFileWriter.init |
2 | 0 | 0 |
meth |
BaseFileWriter.close |
1 | 0 | 0 |
meth |
BaseFileWriter.fileno |
1 | 0 | 0 |
meth |
BaseFileWriter.flush |
1 | 0 | 0 |
meth |
BaseFileWriter.write |
2 | 0 | 0 |
meth |
BaseFileWriter.file_path |
1 | 0 | 0 |
meth |
BaseFileWriter._incr_stats |
3 | 0 | 0 |
meth |
BaseFileWriter._dump_state |
1 | 0 | 0 |
deepspeed.io.base_io_buffer (28 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Base_IO_Buffer.init |
3 | 0 | 0 |
meth |
Base_IO_Buffer.fill |
3 | 0 | 0 |
meth |
Base_IO_Buffer.drain |
4 | 0 | 0 |
meth |
Base_IO_Buffer.is_empty |
1 | 0 | 0 |
meth |
Base_IO_Buffer.is_full |
1 | 0 | 0 |
meth |
Base_IO_Buffer.get_buffer |
1 | 0 | 0 |
meth |
Base_IO_Buffer.get_offset |
1 | 0 | 0 |
meth |
Base_IO_Buffer.get_aligned_num_bytes |
1 | 0 | 0 |
meth |
Base_IO_Buffer.get_unaligned_num_bytes |
1 | 0 | 0 |
meth |
Base_IO_Buffer.reset |
1 | 0 | 0 |
meth |
Base_IO_Buffer.complete_ongoing_drain |
1 | 0 | 0 |
meth |
Base_IO_Buffer._drain |
5 | 0 | 0 |
meth |
Base_IO_Buffer.fill_buffer |
5 | 0 | 0 |
deepspeed.io.double_io_buffer (23 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Double_IO_Buffer.init |
3 | 0 | 0 |
meth |
Double_IO_Buffer.fill |
3 | 0 | 0 |
meth |
Double_IO_Buffer.drain |
4 | 0 | 0 |
meth |
Double_IO_Buffer.get_buffer |
1 | 0 | 0 |
meth |
Double_IO_Buffer.get_offset |
1 | 0 | 0 |
meth |
Double_IO_Buffer.get_aligned_num_bytes |
1 | 0 | 0 |
meth |
Double_IO_Buffer.get_unaligned_num_bytes |
1 | 0 | 0 |
meth |
Double_IO_Buffer.is_full |
1 | 0 | 0 |
meth |
Double_IO_Buffer.is_empty |
1 | 0 | 0 |
meth |
Double_IO_Buffer.reset |
1 | 0 | 0 |
meth |
Double_IO_Buffer.complete_ongoing_drain |
1 | 0 | 0 |
meth |
Double_IO_Buffer._split_buffer |
1 | 0 | 0 |
meth |
Double_IO_Buffer._validate_buffer_index |
2 | 0 | 0 |
meth |
Double_IO_Buffer._wait_for_drain |
1 | 0 | 0 |
meth |
Double_IO_Buffer._is_ongoing_drain |
1 | 0 | 0 |
deepspeed.io.fast_file_writer (43 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FastFileWriter.init |
3 | 0 | 0 |
meth |
FastFileWriter.write |
2 | 0 | 0 |
meth |
FastFileWriter.split_index_list |
3 | 0 | 0 |
meth |
FastFileWriter.save_torch_storage_object_list |
3 | 0 | 0 |
meth |
FastFileWriter.close |
1 | 0 | 0 |
meth |
FastFileWriter.fileno |
1 | 0 | 0 |
meth |
FastFileWriter.flush |
1 | 0 | 0 |
meth |
FastFileWriter.del |
1 | 0 | 0 |
meth |
FastFileWriter._fini |
1 | 0 | 0 |
meth |
FastFileWriter._fill_io_buffer |
3 | 0 | 0 |
meth |
FastFileWriter._drain_io_buffer |
2 | 0 | 0 |
meth |
FastFileWriter._io_buffer_is_full |
1 | 0 | 0 |
meth |
FastFileWriter._io_buffer_is_empty |
1 | 0 | 0 |
meth |
FastFileWriter._force_drain |
1 | 0 | 0 |
meth |
FastFileWriter._unaligned_drain |
2 | 0 | 0 |
meth |
FastFileWriter._dump_state |
1 | 0 | 0 |
meth |
FastFileWriter._update_write_stats |
3 | 0 | 0 |
meth |
FastFileWriter._write_from_tensor |
2 | 0 | 0 |
meth |
FastFileWriter._save_storage_list |
3 | 0 | 0 |
meth |
FastFileWriter._convert_to_byte_tensors |
3 | 0 | 0 |
meth |
FastFileWriter._partition_byte_tensors |
5 | 0 | 0 |
deepspeed.io.mock_file_writer (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MockFileWriter.init |
2 | 0 | 0 |
meth |
MockFileWriter.close |
1 | 0 | 0 |
meth |
MockFileWriter.fileno |
1 | 0 | 0 |
meth |
MockFileWriter.flush |
1 | 0 | 0 |
meth |
MockFileWriter.write |
2 | 0 | 0 |
meth |
MockFileWriter.save_torch_storage_object_list |
3 | 0 | 0 |
meth |
MockFileWriter._save_torch_storage_object |
3 | 0 | 0 |
meth |
MockFileWriter._write |
2 | 0 | 0 |
deepspeed.io.py_file_writer (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PyFileWriter.init |
2 | 0 | 0 |
meth |
PyFileWriter.close |
1 | 0 | 0 |
meth |
PyFileWriter.fileno |
1 | 0 | 0 |
meth |
PyFileWriter.flush |
1 | 0 | 0 |
meth |
PyFileWriter.write |
2 | 0 | 0 |
deepspeed.io.single_io_buffer (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Single_IO_Buffer.init |
3 | 0 | 0 |
meth |
Single_IO_Buffer.fill |
3 | 0 | 0 |
meth |
Single_IO_Buffer.drain |
4 | 0 | 0 |
meth |
Single_IO_Buffer.get_buffer |
1 | 0 | 0 |
meth |
Single_IO_Buffer.get_offset |
1 | 0 | 0 |
meth |
Single_IO_Buffer.get_aligned_num_bytes |
1 | 0 | 0 |
meth |
Single_IO_Buffer.get_unaligned_num_bytes |
1 | 0 | 0 |
meth |
Single_IO_Buffer.is_full |
1 | 0 | 0 |
meth |
Single_IO_Buffer.is_empty |
1 | 0 | 0 |
meth |
Single_IO_Buffer.reset |
1 | 0 | 0 |
deepspeed.io.utils (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
tensor_to_bytes |
2 | 0 | 0 |
func |
required_minimum_torch_version |
3 | 0 | 0 |
func |
obj_serialization_details |
1 | 0 | 0 |
func |
bytes_to_tensor |
2 | 0 | 0 |
deepspeed.launcher.launch (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
terminate_process_tree |
2 | 0 | 0 |
func |
parse_args |
1 | 0 | 0 |
func |
main |
1 | 0 | 0 |
deepspeed.launcher.launcher_helper (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
env_mapping |
4 | 0 | 0 |
func |
parse_args |
2 | 0 | 0 |
func |
main |
2 | 0 | 0 |
deepspeed.launcher.multinode_runner (81 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SlurmRunner.init |
4 | 0 | 0 |
meth |
SlurmRunner.backend_exists |
1 | 0 | 0 |
meth |
SlurmRunner.get_cmd |
3 | 0 | 0 |
prop |
SlurmRunner.name |
1 | 0 | 0 |
attr |
SlurmRunner.resource_pool |
1 | 0 | 0 |
meth |
MVAPICHRunner.init |
4 | 0 | 0 |
meth |
MVAPICHRunner.backend_exists |
1 | 0 | 0 |
meth |
MVAPICHRunner.validate_args |
1 | 0 | 0 |
meth |
MVAPICHRunner.get_cmd |
3 | 0 | 0 |
prop |
MVAPICHRunner.name |
1 | 0 | 0 |
attr |
MVAPICHRunner.resource_pool |
1 | 0 | 0 |
meth |
MultiNodeRunner.init |
3 | 0 | 0 |
meth |
MultiNodeRunner.backend_exists |
1 | 0 | 0 |
meth |
MultiNodeRunner.get_cmd |
3 | 0 | 0 |
meth |
MultiNodeRunner.add_export |
3 | 0 | 0 |
meth |
MultiNodeRunner.parse_user_args |
1 | 0 | 0 |
meth |
MultiNodeRunner.validate_args |
1 | 0 | 0 |
prop |
MultiNodeRunner.name |
1 | 0 | 0 |
attr |
MultiNodeRunner.args |
1 | 0 | 0 |
attr |
MultiNodeRunner.user_arguments |
1 | 0 | 0 |
attr |
MultiNodeRunner.user_script |
1 | 0 | 0 |
attr |
MultiNodeRunner.world_info_base64 |
1 | 0 | 0 |
attr |
MultiNodeRunner.exports |
1 | 0 | 0 |
meth |
PDSHRunner.init |
3 | 0 | 0 |
meth |
PDSHRunner.backend_exists |
1 | 0 | 0 |
meth |
PDSHRunner.parse_user_args |
1 | 0 | 0 |
meth |
PDSHRunner.get_cmd |
3 | 0 | 0 |
prop |
PDSHRunner.name |
1 | 0 | 0 |
meth |
MPICHRunner.init |
4 | 0 | 0 |
meth |
MPICHRunner.backend_exists |
1 | 0 | 0 |
meth |
MPICHRunner.validate_args |
1 | 0 | 0 |
meth |
MPICHRunner.get_cmd |
3 | 0 | 0 |
prop |
MPICHRunner.name |
1 | 0 | 0 |
attr |
MPICHRunner.resource_pool |
1 | 0 | 0 |
meth |
OpenMPIRunner.init |
4 | 0 | 0 |
meth |
OpenMPIRunner.backend_exists |
1 | 0 | 0 |
meth |
OpenMPIRunner.validate_args |
1 | 0 | 0 |
meth |
OpenMPIRunner.get_cmd |
3 | 0 | 0 |
prop |
OpenMPIRunner.name |
1 | 0 | 0 |
attr |
OpenMPIRunner.resource_pool |
1 | 0 | 0 |
meth |
IMPIRunner.init |
4 | 0 | 0 |
meth |
IMPIRunner.backend_exists |
1 | 0 | 0 |
meth |
IMPIRunner.validate_args |
1 | 0 | 0 |
meth |
IMPIRunner.get_cmd |
3 | 0 | 0 |
prop |
IMPIRunner.name |
1 | 0 | 0 |
attr |
IMPIRunner.resource_pool |
1 | 0 | 0 |
deepspeed.launcher.runner (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
fetch_hostfile |
2 | 0 | 0 |
attr |
DEEPSPEED_ENVIRONMENT_PATHS |
1 | 0 | 0 |
func |
parse_args |
2 | 0 | 0 |
func |
parse_resource_filter |
4 | 0 | 0 |
func |
parse_num_nodes |
3 | 2 | 0 |
attr |
DEEPSPEED_ENVIRONMENT_NAME |
1 | 0 | 0 |
func |
encode_world_info |
2 | 0 | 0 |
func |
parse_inclusion_exclusion |
4 | 0 | 0 |
func |
run_autotuning |
3 | 0 | 0 |
func |
main |
2 | 0 | 0 |
deepspeed.linear.context_manager (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Init.init |
3 | 0 | 0 |
meth |
Init.enter |
1 | 0 | 0 |
meth |
Init.exit |
3 | 0 | 0 |
attr |
Init.lora_config |
1 | 0 | 0 |
attr |
Init.quant_config |
1 | 0 | 0 |
func |
init_lora |
2 | 0 | 0 |
deepspeed.linear.optimized_linear (36 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OptimizedLinear.new |
9 | 5 | 0 |
meth |
LoRAOptimizedLinear.init |
9 | 5 | 0 |
meth |
LoRAOptimizedLinear.disable |
1 | 0 | 0 |
meth |
LoRAOptimizedLinear.init_lora |
1 | 0 | 0 |
meth |
LoRAOptimizedLinear._load_from_state_dict |
8 | 0 | 0 |
meth |
LoRAOptimizedLinear.full_weight |
1 | 0 | 0 |
meth |
LoRAOptimizedLinear.linear_without_F_linear |
3 | 0 | 0 |
meth |
LoRAOptimizedLinear.forward |
2 | 0 | 0 |
attr |
LoRAOptimizedLinear.input_dim |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.output_dim |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.bias |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.lora_config |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.quantization_config |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.device |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.linear_cls |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.dtype |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.zero_shards |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.sharded_weight_size |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.disabled |
1 | 0 | 0 |
attr |
LoRAOptimizedLinear.weight |
1 | 0 | 0 |
deepspeed.linear.quantization (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
QuantizedParameter.new |
5 | 4 | 0 |
meth |
QuantizedParameter._ensure_quantized |
2 | 1 | 0 |
meth |
QuantizedParameter.offload |
2 | 0 | 0 |
meth |
QuantizedParameter.getstate |
1 | 0 | 0 |
meth |
QuantizedParameter.setstate |
2 | 0 | 0 |
meth |
QuantizedParameter.deepcopy |
2 | 0 | 0 |
meth |
QuantizedParameter.copy |
1 | 0 | 0 |
meth |
QuantizedParameter.cuda |
3 | 0 | 0 |
meth |
QuantizedParameter.to |
3 | 0 | 0 |
meth |
QuantizedLinear.init |
6 | 4 | 0 |
attr |
QuantizedLinear.weight |
1 | 0 | 0 |
deepspeed.model_implementations.diffusers.unet (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSUNet.init |
3 | 0 | 0 |
meth |
DSUNet._graph_replay |
3 | 0 | 0 |
meth |
DSUNet.forward |
3 | 0 | 0 |
meth |
DSUNet._create_cuda_graph |
3 | 0 | 0 |
meth |
DSUNet._forward |
8 | 0 | 0 |
attr |
DSUNet.unet |
1 | 0 | 0 |
attr |
DSUNet.in_channels |
1 | 0 | 0 |
attr |
DSUNet.device |
1 | 0 | 0 |
attr |
DSUNet.dtype |
1 | 0 | 0 |
attr |
DSUNet.config |
1 | 0 | 0 |
attr |
DSUNet.fwd_count |
1 | 0 | 0 |
attr |
DSUNet.cuda_graph_created |
1 | 0 | 0 |
deepspeed.model_implementations.diffusers.vae (49 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSVAE.init |
3 | 0 | 0 |
meth |
DSVAE._graph_replay_decoder |
3 | 0 | 0 |
meth |
DSVAE._decode |
4 | 0 | 0 |
meth |
DSVAE._create_cuda_graph_decoder |
3 | 0 | 0 |
meth |
DSVAE.decode |
3 | 0 | 0 |
meth |
DSVAE._graph_replay_encoder |
3 | 0 | 0 |
meth |
DSVAE._encode |
3 | 0 | 0 |
meth |
DSVAE._create_cuda_graph_encoder |
3 | 0 | 0 |
meth |
DSVAE.encode |
3 | 0 | 0 |
meth |
DSVAE._graph_replay |
3 | 0 | 0 |
meth |
DSVAE.forward |
3 | 0 | 0 |
meth |
DSVAE._create_cuda_graph |
3 | 0 | 0 |
meth |
DSVAE._forward |
5 | 0 | 0 |
attr |
DSVAE.vae |
1 | 0 | 0 |
attr |
DSVAE.config |
1 | 0 | 0 |
attr |
DSVAE.device |
1 | 0 | 0 |
attr |
DSVAE.dtype |
1 | 0 | 0 |
attr |
DSVAE.decoder_cuda_graph_created |
1 | 0 | 0 |
attr |
DSVAE.encoder_cuda_graph_created |
1 | 0 | 0 |
attr |
DSVAE.all_cuda_graph_created |
1 | 0 | 0 |
deepspeed.model_implementations.features.cuda_graph (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CUDAGraph.init |
2 | 0 | 0 |
meth |
CUDAGraph._create_cuda_graph |
1 | 0 | 0 |
meth |
CUDAGraph._graph_replay |
1 | 0 | 0 |
attr |
CUDAGraph.enable_cuda_graph |
1 | 0 | 0 |
deepspeed.model_implementations.transformers.clip_encoder (28 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DSClipEncoder.init |
3 | 0 | 0 |
meth |
DSClipEncoder._build_causal_attention_mask |
4 | 0 | 0 |
meth |
DSClipEncoder._graph_replay |
3 | 0 | 0 |
meth |
DSClipEncoder.forward |
3 | 0 | 0 |
meth |
DSClipEncoder._create_cuda_graph |
3 | 0 | 0 |
meth |
DSClipEncoder._forward |
3 | 0 | 0 |
attr |
DSClipEncoder.enc |
1 | 0 | 0 |
attr |
DSClipEncoder.device |
1 | 0 | 0 |
attr |
DSClipEncoder.dtype |
1 | 0 | 0 |
attr |
DSClipEncoder.cuda_graph_created |
1 | 0 | 0 |
attr |
DSClipEncoder.static_inputs |
1 | 0 | 0 |
attr |
DSClipEncoder.static_kwargs |
1 | 0 | 0 |
attr |
DSClipEncoder.static_output |
1 | 0 | 0 |
attr |
DSClipEncoder.iter |
1 | 0 | 0 |
attr |
DSClipEncoder.config |
1 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_base (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedTransformerBase.init |
1 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_bert (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedBERTInference.init |
7 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_bloom (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedBloomInference.init |
7 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_gpt (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedGPTInference.init |
7 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_llama2 (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedLlama2Inference.init |
7 | 0 | 0 |
meth |
DeepSpeedLlama2Inference.forward |
3 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_megatron_gpt (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedMegatronGPTInference.init |
7 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_opt (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedOPTInference.init |
7 | 0 | 0 |
deepspeed.model_implementations.transformers.ds_transformer (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedTransformerInference.init |
7 | 0 | 0 |
meth |
DeepSpeedTransformerInference.allocate_workspace |
2 | 0 | 0 |
meth |
DeepSpeedTransformerInference.reset_cache |
1 | 0 | 0 |
meth |
DeepSpeedTransformerInference.forward |
20 | 0 | 0 |
attr |
DeepSpeedTransformerInference.config |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.layer_past |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.layer_norm |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.attention |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.mlp |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.norm_w |
1 | 0 | 0 |
attr |
DeepSpeedTransformerInference.norm_b |
1 | 0 | 0 |
deepspeed.module_inject.auto_tp (89 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Loading.is_load_module |
1 | 0 | 0 |
meth |
Loading.load_buffer |
3 | 0 | 0 |
meth |
Loading.load |
4 | 0 | 0 |
func |
move |
4 | 0 | 0 |
meth |
AutoTP.init |
9 | 1 | 0 |
meth |
AutoTP.in_module_list |
2 | 0 | 0 |
meth |
AutoTP.get_module_list |
1 | 0 | 0 |
meth |
AutoTP.supported |
1 | 0 | 0 |
meth |
AutoTP.get_layers |
2 | 0 | 0 |
meth |
AutoTP.update_policy_list |
3 | 0 | 0 |
meth |
AutoTP.kernel_supported |
1 | 0 | 0 |
meth |
AutoTP.tp_parser |
1 | 0 | 0 |
meth |
AutoTP.set_tensor_parallel_config |
3 | 0 | 0 |
meth |
AutoTP._replace |
4 | 0 | 0 |
meth |
AutoTP._replace_with_config |
3 | 0 | 0 |
meth |
AutoTP._create_row_parallel_layer |
4 | 2 | 0 |
meth |
AutoTP._create_column_parallel_layer |
4 | 2 | 0 |
meth |
AutoTP._slice_embedding |
4 | 0 | 0 |
meth |
AutoTP.update_mp_params |
2 | 0 | 0 |
meth |
AutoTP.update_linear_policies |
1 | 0 | 0 |
meth |
AutoTP._replace_module |
4 | 0 | 0 |
meth |
AutoTP.get_model_num_kv_heads |
2 | 0 | 0 |
meth |
AutoTP._replace_last_linear_module |
2 | 0 | 0 |
attr |
AutoTP.module |
1 | 0 | 0 |
attr |
AutoTP.all_reduce_linears |
1 | 0 | 0 |
attr |
AutoTP.prefix |
1 | 0 | 0 |
attr |
AutoTP.state_dict |
1 | 0 | 0 |
attr |
AutoTP.mp_size |
1 | 0 | 0 |
attr |
AutoTP.mp_group |
1 | 0 | 0 |
attr |
AutoTP.linear_layer_setting |
1 | 0 | 0 |
attr |
AutoTP.orig_layer_impl |
1 | 0 | 0 |
attr |
AutoTP.linear_policies |
1 | 0 | 0 |
attr |
AutoTP.conv_linear_layer |
1 | 0 | 0 |
attr |
AutoTP.partition_config |
1 | 0 | 0 |
meth |
ReplaceWithTensorSlicing.init |
5 | 0 | 0 |
meth |
ReplaceWithTensorSlicing.merge_assert |
3 | 0 | 0 |
meth |
ReplaceWithTensorSlicing.strided_copy |
6 | 5 | 0 |
meth |
ReplaceWithTensorSlicing.copy |
5 | 0 | 0 |
attr |
ReplaceWithTensorSlicing.out_dim |
1 | 0 | 0 |
attr |
ReplaceWithTensorSlicing.in_dim |
1 | 0 | 0 |
attr |
ReplaceWithTensorSlicing.mp_size |
1 | 0 | 0 |
attr |
ReplaceWithTensorSlicing.gpu_index |
1 | 0 | 0 |
deepspeed.module_inject.auto_tp_model_utils (13 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
build_mpt_atten_bias_tensor |
7 | 3 | 0 |
func |
build_mpt_alibi_tensor |
6 | 1 | 0 |
func |
get_alibi_mask |
4 | 0 | 0 |
deepspeed.module_inject.autotp_config (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TPLayerSpec.post_init |
1 | 0 | 0 |
meth |
TPLayerSpec._normalize_shape |
2 | 0 | 0 |
meth |
TPLayerSpec._validate_shape_format |
1 | 0 | 0 |
meth |
AutoTPConfig._convert_shape |
2 | 0 | 0 |
deepspeed.module_inject.containers.base (107 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BaseTransformerContainer.init |
6 | 0 | 0 |
meth |
BaseTransformerContainer.create_ds_model_config |
1 | 0 | 0 |
meth |
BaseTransformerContainer.check_meta_tensor_support |
1 | 0 | 0 |
meth |
BaseTransformerContainer.initialize_tensors |
2 | 0 | 0 |
meth |
BaseTransformerContainer.convert_to_required_dtype |
1 | 0 | 0 |
meth |
BaseTransformerContainer.get_rotary_dim |
1 | 0 | 0 |
meth |
BaseTransformerContainer.set_moe |
2 | 0 | 0 |
meth |
BaseTransformerContainer.set_tensor_parallel_config |
3 | 0 | 0 |
meth |
BaseTransformerContainer.set_quantization_config |
2 | 0 | 0 |
meth |
BaseTransformerContainer.set_hidden_heads |
5 | 0 | 0 |
meth |
BaseTransformerContainer.set_attention |
5 | 0 | 0 |
meth |
BaseTransformerContainer.set_mlp |
5 | 0 | 0 |
meth |
BaseTransformerContainer.set_layernorm |
5 | 0 | 0 |
meth |
BaseTransformerContainer.apply_weight_quantization |
1 | 0 | 0 |
meth |
BaseTransformerContainer.attention_quantization |
1 | 0 | 0 |
meth |
BaseTransformerContainer.mlp_quantization |
1 | 0 | 0 |
meth |
BaseTransformerContainer.apply_tensor_parallelism |
2 | 0 | 0 |
meth |
BaseTransformerContainer.attention_qkv_mp |
3 | 0 | 0 |
meth |
BaseTransformerContainer.attention_o_mp |
3 | 0 | 0 |
meth |
BaseTransformerContainer.mlp_inter_mp |
3 | 0 | 0 |
meth |
BaseTransformerContainer.mlp_output_mp |
3 | 0 | 0 |
meth |
BaseTransformerContainer.copy_data_to_new_module |
1 | 0 | 0 |
meth |
BaseTransformerContainer.transpose |
1 | 0 | 0 |
meth |
BaseTransformerContainer.transpose_attention |
1 | 0 | 0 |
meth |
BaseTransformerContainer.transpose_mlp |
1 | 0 | 0 |
meth |
BaseTransformerContainer.transpose_impl |
2 | 0 | 0 |
meth |
BaseTransformerContainer.get_all_params |
1 | 0 | 0 |
meth |
BaseTransformerContainer.get_attn_params |
1 | 0 | 0 |
meth |
BaseTransformerContainer.get_mlp_params |
1 | 0 | 0 |
attr |
BaseTransformerContainer.policy |
1 | 0 | 0 |
attr |
BaseTransformerContainer.config |
1 | 0 | 0 |
attr |
BaseTransformerContainer.model_config |
1 | 0 | 0 |
attr |
BaseTransformerContainer.layer_id |
1 | 0 | 0 |
attr |
BaseTransformerContainer.child |
1 | 0 | 0 |
attr |
BaseTransformerContainer.megatron_v2 |
1 | 0 | 0 |
attr |
BaseTransformerContainer.scale_attention |
1 | 0 | 0 |
attr |
BaseTransformerContainer.ckpt_load_enabled |
1 | 0 | 0 |
attr |
BaseTransformerContainer.hidden_size |
1 | 0 | 0 |
attr |
BaseTransformerContainer.intermediate_size |
1 | 0 | 0 |
attr |
BaseTransformerContainer.num_attention_heads |
1 | 0 | 0 |
attr |
BaseTransformerContainer.mp_size |
1 | 0 | 0 |
attr |
BaseTransformerContainer.pre_layer_norm |
1 | 0 | 0 |
attr |
BaseTransformerContainer.dtype |
1 | 0 | 0 |
attr |
BaseTransformerContainer.attn_linear_layer |
1 | 0 | 0 |
attr |
BaseTransformerContainer.mlp_linear_layer |
1 | 0 | 0 |
attr |
BaseTransformerContainer.return_tuple |
1 | 0 | 0 |
attr |
BaseTransformerContainer.triangular_masking |
1 | 0 | 0 |
attr |
BaseTransformerContainer.local_attention |
1 | 0 | 0 |
attr |
BaseTransformerContainer.window_size |
1 | 0 | 0 |
attr |
BaseTransformerContainer.mlp_act_func_type |
1 | 0 | 0 |
attr |
BaseTransformerContainer.norm_type |
1 | 0 | 0 |
attr |
BaseTransformerContainer.training_mp_size |
1 | 0 | 0 |
attr |
BaseTransformerContainer.bigscience_bloom |
1 | 0 | 0 |
attr |
BaseTransformerContainer.max_out_tokens |
1 | 0 | 0 |
attr |
BaseTransformerContainer.min_out_tokens |
1 | 0 | 0 |
attr |
BaseTransformerContainer.scale_attn_by_inverse_layer_idx |
1 | 0 | 0 |
attr |
BaseTransformerContainer.use_mup |
1 | 0 | 0 |
attr |
BaseTransformerContainer.return_single_tuple |
1 | 0 | 0 |
attr |
BaseTransformerContainer.rotary_dim |
1 | 0 | 0 |
attr |
BaseTransformerContainer.mlp_after_attn |
1 | 0 | 0 |
attr |
BaseTransformerContainer.qkvw |
1 | 0 | 0 |
attr |
BaseTransformerContainer.qkvb |
1 | 0 | 0 |
attr |
BaseTransformerContainer.dense_w |
1 | 0 | 0 |
attr |
BaseTransformerContainer.dense_b |
1 | 0 | 0 |
attr |
BaseTransformerContainer.attn_nw |
1 | 0 | 0 |
attr |
BaseTransformerContainer.attn_nb |
1 | 0 | 0 |
attr |
BaseTransformerContainer.input_nw |
1 | 0 | 0 |
attr |
BaseTransformerContainer.input_nb |
1 | 0 | 0 |
attr |
BaseTransformerContainer.mp_group |
1 | 0 | 0 |
attr |
BaseTransformerContainer.use_triton |
1 | 0 | 0 |
meth |
BaseConvolutionContainer.init |
1 | 0 | 0 |
deepspeed.module_inject.containers.base_moe (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BaseTransformerMoEContainer.init |
2 | 0 | 0 |
meth |
BaseTransformerMoEContainer.create_ds_model_config |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.initialize_tensors |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.set_mlp |
2 | 0 | 0 |
meth |
BaseTransformerMoEContainer.transpose |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.transpose_mlp |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.transpose_residual |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.apply_tensor_parallelism |
2 | 0 | 0 |
meth |
BaseTransformerMoEContainer.mlp_mp |
1 | 0 | 0 |
meth |
BaseTransformerMoEContainer.copy_data_to_new_module |
1 | 0 | 0 |
attr |
BaseTransformerMoEContainer.num_experts |
1 | 0 | 0 |
attr |
BaseTransformerMoEContainer.ep_world_size |
1 | 0 | 0 |
attr |
BaseTransformerMoEContainer.local_ep_size |
1 | 0 | 0 |
attr |
BaseTransformerMoEContainer.layer_norm_eps |
1 | 0 | 0 |
deepspeed.module_inject.containers.bert (18 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HFBertLayerPolicy.init |
3 | 0 | 0 |
meth |
HFBertLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFBertLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFBertLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFBertLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFBertLayerPolicy.client_module |
1 | 0 | 0 |
attr |
HFBertLayerPolicy.cuda_graph_supported |
1 | 0 | 0 |
meth |
DS_BERTContainer.init |
2 | 0 | 0 |
meth |
DS_BERTContainer.create_module |
2 | 0 | 0 |
attr |
DS_BERTContainer.return_tuple |
1 | 0 | 0 |
attr |
DS_BERTContainer.triangular_masking |
1 | 0 | 0 |
attr |
DS_BERTContainer.use_triton |
1 | 0 | 0 |
deepspeed.module_inject.containers.bloom (29 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BLOOMLayerPolicy.init |
5 | 0 | 0 |
meth |
BLOOMLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
BLOOMLayerPolicy.attention |
2 | 0 | 0 |
meth |
BLOOMLayerPolicy.mlp |
2 | 0 | 0 |
meth |
BLOOMLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
BLOOMLayerPolicy.client_module |
1 | 0 | 0 |
meth |
DS_BloomContainer.init |
2 | 0 | 0 |
meth |
DS_BloomContainer.create_module |
2 | 0 | 0 |
meth |
DS_BloomContainer.attention_qkv_mp |
3 | 0 | 0 |
meth |
DS_BloomContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_BloomContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_BloomContainer.load_params |
6 | 0 | 0 |
attr |
DS_BloomContainer.bigscience_bloom |
1 | 0 | 0 |
attr |
DS_BloomContainer.triangular_masking |
1 | 0 | 0 |
deepspeed.module_inject.containers.clip (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HFCLIPLayerPolicy.init |
3 | 0 | 0 |
meth |
HFCLIPLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFCLIPLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFCLIPLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFCLIPLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFCLIPLayerPolicy.client_module |
1 | 0 | 0 |
attr |
HFCLIPLayerPolicy.cuda_graph_supported |
1 | 0 | 0 |
meth |
DS_CLIPContainer.init |
2 | 0 | 0 |
meth |
DS_CLIPContainer.create_module |
2 | 0 | 0 |
deepspeed.module_inject.containers.distil_bert (20 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_DistilBERTContainer.init |
2 | 0 | 0 |
meth |
DS_DistilBERTContainer.create_module |
2 | 0 | 0 |
attr |
DS_DistilBERTContainer.triangular_masking |
1 | 0 | 0 |
attr |
DS_DistilBERTContainer.return_single_tuple |
1 | 0 | 0 |
attr |
DS_DistilBERTContainer.use_triton |
1 | 0 | 0 |
meth |
HFDistilBertLayerPolicy.init |
4 | 0 | 0 |
meth |
HFDistilBertLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFDistilBertLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFDistilBertLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFDistilBertLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFDistilBertLayerPolicy.client_module |
1 | 0 | 0 |
attr |
HFDistilBertLayerPolicy.preln |
1 | 0 | 0 |
attr |
HFDistilBertLayerPolicy.cuda_graph_supported |
1 | 0 | 0 |
deepspeed.module_inject.containers.features.gated_mlp (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HybridGatedMLPContainer.set_mlp |
5 | 0 | 0 |
meth |
HybridGatedMLPContainer.set_mlp_gate |
1 | 0 | 0 |
meth |
HybridGatedMLPContainer.mlp_inter_mp |
3 | 0 | 0 |
meth |
HybridGatedMLPContainer.release_mlp |
1 | 0 | 0 |
meth |
HybridGatedMLPContainer.reset_mlp |
1 | 0 | 0 |
meth |
HybridGatedMLPContainer.set_mlp_params_wo_copy |
2 | 0 | 0 |
meth |
HybridGatedMLPContainer.get_mlp_params |
1 | 0 | 0 |
deepspeed.module_inject.containers.features.hybrid_engine (25 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HybridEngineContainer.initialize_tensors |
2 | 0 | 0 |
meth |
HybridEngineContainer.transform_for_training |
1 | 0 | 0 |
meth |
HybridEngineContainer.transform_for_inference |
1 | 0 | 0 |
meth |
HybridEngineContainer.set_lora_params |
1 | 0 | 0 |
meth |
HybridEngineContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
HybridEngineContainer.fuse_lora |
1 | 0 | 0 |
meth |
HybridEngineContainer.unfuse_lora |
1 | 0 | 0 |
meth |
HybridEngineContainer.apply_tensor_parallelism |
3 | 0 | 0 |
meth |
HybridEngineContainer._release_params |
2 | 1 | 0 |
meth |
HybridEngineContainer.release_memory |
1 | 0 | 0 |
meth |
HybridEngineContainer.release_qkv |
1 | 0 | 0 |
meth |
HybridEngineContainer.release_mlp |
1 | 0 | 0 |
meth |
HybridEngineContainer.reset_params |
1 | 0 | 0 |
meth |
HybridEngineContainer.reset_qkv |
1 | 0 | 0 |
meth |
HybridEngineContainer.reset_mlp |
1 | 0 | 0 |
meth |
HybridEngineContainer.get_lora_params |
1 | 0 | 0 |
meth |
HybridEngineContainer.set_params_wo_copy |
2 | 0 | 0 |
meth |
HybridEngineContainer.set_attn_params_wo_copy |
2 | 0 | 0 |
meth |
HybridEngineContainer.set_mlp_params_wo_copy |
2 | 0 | 0 |
deepspeed.module_inject.containers.features.hybrid_megatron (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HybridMegatronContainer._align_qkv |
2 | 1 | 0 |
meth |
HybridMegatronContainer._partition_qkv |
2 | 1 | 0 |
meth |
HybridMegatronContainer.transform_for_training |
1 | 0 | 0 |
deepspeed.module_inject.containers.features.megatron (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MegatronContainer.init |
2 | 0 | 0 |
meth |
MegatronContainer._align_qkv_transposed |
2 | 0 | 0 |
meth |
MegatronContainer.transpose |
1 | 0 | 0 |
attr |
MegatronContainer.megatron_v2 |
1 | 0 | 0 |
deepspeed.module_inject.containers.features.meta_tensor (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MetaTensorContainer.init |
2 | 0 | 0 |
meth |
MetaTensorContainer.initialize_tensors |
2 | 0 | 0 |
meth |
MetaTensorContainer.apply_tensor_parallelism |
3 | 0 | 0 |
meth |
MetaTensorContainer.copy_data_to_new_module |
1 | 0 | 0 |
meth |
MetaTensorContainer.transpose |
1 | 0 | 0 |
meth |
MetaTensorContainer.load_params |
6 | 0 | 0 |
attr |
MetaTensorContainer.is_meta |
1 | 0 | 0 |
attr |
MetaTensorContainer.ckpt_load_enabled |
1 | 0 | 0 |
deepspeed.module_inject.containers.features.split_qkv (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HybridSplitQKVContainer.set_attention |
5 | 0 | 0 |
meth |
HybridSplitQKVContainer.set_q_k_v |
1 | 0 | 0 |
meth |
HybridSplitQKVContainer.attention_qkv_mp |
3 | 0 | 0 |
meth |
HybridSplitQKVContainer.release_qkv |
1 | 0 | 0 |
meth |
HybridSplitQKVContainer.reset_qkv |
1 | 0 | 0 |
meth |
HybridSplitQKVContainer.reset_qkv_experimental |
1 | 0 | 0 |
meth |
HybridSplitQKVContainer.set_attn_params_wo_copy |
2 | 0 | 0 |
meth |
HybridSplitQKVContainer.get_attn_params |
1 | 0 | 0 |
deepspeed.module_inject.containers.gpt2 (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HFGPT2LayerPolicy.init |
3 | 0 | 0 |
meth |
HFGPT2LayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFGPT2LayerPolicy.attention |
2 | 0 | 0 |
meth |
HFGPT2LayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFGPT2LayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFGPT2LayerPolicy.client_module |
1 | 0 | 0 |
meth |
DS_GPT2Container.init |
2 | 0 | 0 |
meth |
DS_GPT2Container.create_module |
2 | 0 | 0 |
deepspeed.module_inject.containers.gptj (23 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_GPTJContainer.init |
2 | 0 | 0 |
meth |
DS_GPTJContainer.create_module |
2 | 0 | 0 |
meth |
DS_GPTJContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_GPTJContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_GPTJContainer.set_q_k_v |
1 | 0 | 0 |
meth |
DS_GPTJContainer.load_params |
6 | 0 | 0 |
meth |
HFGPTJLayerPolicy.init |
3 | 0 | 0 |
meth |
HFGPTJLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFGPTJLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFGPTJLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFGPTJLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFGPTJLayerPolicy.client_module |
1 | 0 | 0 |
deepspeed.module_inject.containers.gptneo (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_GPTNEOContainer.init |
2 | 0 | 0 |
meth |
DS_GPTNEOContainer.create_module |
2 | 0 | 0 |
meth |
DS_GPTNEOContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_GPTNEOContainer.set_q_k_v |
1 | 0 | 0 |
meth |
DS_GPTNEOContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_GPTNEOContainer.load_params |
6 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.init |
3 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.get_q_k_v |
1 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFGPTNEOLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFGPTNEOLayerPolicy.client_module |
1 | 0 | 0 |
deepspeed.module_inject.containers.gptneox (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GPTNEOXLayerPolicy.init |
5 | 0 | 0 |
meth |
GPTNEOXLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
GPTNEOXLayerPolicy.attention |
2 | 0 | 0 |
meth |
GPTNEOXLayerPolicy.mlp |
2 | 0 | 0 |
meth |
GPTNEOXLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
GPTNEOXLayerPolicy.client_module |
1 | 0 | 0 |
meth |
DS_GPTNEOXContainer.init |
2 | 0 | 0 |
meth |
DS_GPTNEOXContainer.create_module |
2 | 0 | 0 |
meth |
DS_GPTNEOXContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_GPTNEOXContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_GPTNEOXContainer.load_params |
6 | 0 | 0 |
deepspeed.module_inject.containers.internlm (25 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_InternLMContainer.init |
2 | 0 | 0 |
meth |
DS_InternLMContainer.create_module |
2 | 0 | 0 |
meth |
DS_InternLMContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_InternLMContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_InternLMContainer.set_q_k_v |
1 | 0 | 0 |
meth |
DS_InternLMContainer.set_mlp_gate |
1 | 0 | 0 |
meth |
DS_InternLMContainer.load_params |
6 | 0 | 0 |
meth |
InternLMLayerPolicy.init |
3 | 0 | 0 |
meth |
InternLMLayerPolicy._init_orig_layer_class_once |
1 | 0 | 0 |
meth |
InternLMLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
InternLMLayerPolicy.attention |
2 | 0 | 0 |
meth |
InternLMLayerPolicy.mlp |
2 | 0 | 0 |
meth |
InternLMLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
InternLMLayerPolicy.client_module |
1 | 0 | 0 |
deepspeed.module_inject.containers.llama (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
LLAMALayerPolicy.init |
3 | 0 | 0 |
meth |
LLAMALayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
LLAMALayerPolicy.attention |
2 | 0 | 0 |
meth |
LLAMALayerPolicy.mlp |
2 | 0 | 0 |
meth |
LLAMALayerPolicy.layernorm |
1 | 0 | 0 |
attr |
LLAMALayerPolicy.client_module |
1 | 0 | 0 |
meth |
DS_LLAMAContainer.init |
2 | 0 | 0 |
meth |
DS_LLAMAContainer.create_module |
2 | 0 | 0 |
meth |
DS_LLAMAContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_LLAMAContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_LLAMAContainer.set_q_k_v |
1 | 0 | 0 |
meth |
DS_LLAMAContainer.set_mlp_gate |
1 | 0 | 0 |
meth |
DS_LLAMAContainer.load_params |
6 | 0 | 0 |
deepspeed.module_inject.containers.llama2 (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_LLAMA2Container.init |
2 | 0 | 0 |
meth |
DS_LLAMA2Container.create_module |
2 | 0 | 0 |
meth |
DS_LLAMA2Container.set_lora_params |
1 | 0 | 0 |
meth |
DS_LLAMA2Container.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_LLAMA2Container.set_q_k_v |
1 | 0 | 0 |
meth |
DS_LLAMA2Container.set_mlp_gate |
1 | 0 | 0 |
meth |
DS_LLAMA2Container.load_params |
6 | 0 | 0 |
meth |
LLAMA2LayerPolicy.init |
3 | 0 | 0 |
meth |
LLAMA2LayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
LLAMA2LayerPolicy.attention |
2 | 0 | 0 |
meth |
LLAMA2LayerPolicy.mlp |
2 | 0 | 0 |
meth |
LLAMA2LayerPolicy.layernorm |
1 | 0 | 0 |
attr |
LLAMA2LayerPolicy.client_module |
1 | 0 | 0 |
deepspeed.module_inject.containers.megatron_gpt (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MegatronLayerPolicy.init |
3 | 0 | 0 |
meth |
MegatronLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
MegatronLayerPolicy.attention |
2 | 0 | 0 |
meth |
MegatronLayerPolicy.mlp |
3 | 0 | 0 |
meth |
MegatronLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
MegatronLayerPolicy.client_module |
1 | 0 | 0 |
meth |
DS_MegatronGPTContainer.init |
2 | 0 | 0 |
meth |
DS_MegatronGPTContainer.create_module |
2 | 0 | 0 |
deepspeed.module_inject.containers.megatron_gpt_moe (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_MegatronGPTMoEContainer.init |
5 | 0 | 0 |
meth |
DS_MegatronGPTMoEContainer.create_module |
2 | 0 | 0 |
meth |
MegatronMoELayerPolicy.init |
3 | 0 | 0 |
meth |
MegatronMoELayerPolicy.get_num_experts |
1 | 0 | 0 |
meth |
MegatronMoELayerPolicy.mlp |
3 | 0 | 0 |
attr |
MegatronMoELayerPolicy.client_module |
1 | 0 | 0 |
deepspeed.module_inject.containers.opt (25 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DS_OPTContainer.init |
2 | 0 | 0 |
meth |
DS_OPTContainer.create_module |
2 | 0 | 0 |
meth |
DS_OPTContainer.set_lora_params |
1 | 0 | 0 |
meth |
DS_OPTContainer.set_q_k_v |
1 | 0 | 0 |
meth |
DS_OPTContainer.get_lora_matched_pair |
1 | 0 | 0 |
meth |
DS_OPTContainer.load_params |
6 | 0 | 0 |
meth |
HFOPTLayerPolicy.init |
4 | 0 | 0 |
meth |
HFOPTLayerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
HFOPTLayerPolicy.attention |
2 | 0 | 0 |
meth |
HFOPTLayerPolicy.mlp |
2 | 0 | 0 |
meth |
HFOPTLayerPolicy.layernorm |
1 | 0 | 0 |
attr |
HFOPTLayerPolicy.client_module |
1 | 0 | 0 |
attr |
HFOPTLayerPolicy.mlp_act_func_type |
1 | 0 | 0 |
deepspeed.module_inject.containers.unet (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
UNetPolicy.init |
1 | 0 | 0 |
meth |
UNetPolicy.match |
2 | 0 | 0 |
meth |
UNetPolicy.match_replaced |
2 | 0 | 0 |
meth |
UNetPolicy.apply |
3 | 0 | 0 |
meth |
UNetPolicy.attention |
2 | 0 | 0 |
deepspeed.module_inject.containers.vae (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
VAEPolicy.init |
1 | 0 | 0 |
meth |
VAEPolicy.match |
2 | 0 | 0 |
meth |
VAEPolicy.match_replaced |
2 | 0 | 0 |
meth |
VAEPolicy.apply |
3 | 0 | 0 |
meth |
VAEPolicy.attention |
2 | 0 | 0 |
deepspeed.module_inject.fusedqkv_utils (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
shard_chunk_mlp |
5 | 0 | 0 |
func |
prepare_tp_fused_qkvw |
5 | 0 | 0 |
func |
shard_value_with_share_qk |
6 | 0 | 0 |
func |
require_tp_fused_qkvw |
3 | 0 | 0 |
func |
split_by_qkvlist_and_refuse |
5 | 0 | 0 |
deepspeed.module_inject.inject (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
module_inject |
9 | 0 | 0 |
func |
test_hi |
1 | 0 | 0 |
deepspeed.module_inject.layers (112 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GateUpPack_LinearLayer._tp_partition |
2 | 0 | 0 |
meth |
LinearLayer.init |
5 | 0 | 0 |
meth |
LinearLayer.forward |
2 | 0 | 0 |
meth |
LinearLayer.gather_params |
2 | 0 | 0 |
meth |
LinearLayer._tp_partition |
2 | 0 | 0 |
meth |
LinearLayer.uneven_partition |
2 | 0 | 0 |
meth |
LinearLayer.from_weights |
5 | 0 | 0 |
attr |
LinearLayer.weight |
1 | 0 | 0 |
attr |
LinearLayer.bias |
1 | 0 | 0 |
meth |
TensorParallel_Layer.init |
3 | 2 | 1 |
meth |
TensorParallel_Layer.set_keep_module_on_host |
2 | 1 | 0 |
meth |
TensorParallel_Layer.forward |
2 | 0 | 0 |
meth |
TensorParallel_Layer.gather_params |
2 | 0 | 0 |
meth |
TensorParallel_Layer._tp_partition |
2 | 1 | 0 |
meth |
TensorParallel_Layer.config_requires_grad |
2 | 0 | 0 |
meth |
TensorParallel_Layer.config_tp_params |
2 | 0 | 0 |
meth |
TensorParallel_Layer.is_training_mode |
1 | 0 | 0 |
meth |
TensorParallel_Layer.deepcopy |
2 | 0 | 0 |
meth |
TensorParallel_Layer.extra_repr |
1 | 0 | 0 |
meth |
TensorParallel_Layer.move |
2 | 0 | 0 |
attr |
TensorParallel_Layer.mp_group |
1 | 0 | 0 |
attr |
TensorParallel_Layer.world_size |
1 | 0 | 0 |
attr |
TensorParallel_Layer.rank |
1 | 0 | 0 |
attr |
TensorParallel_Layer.name |
1 | 0 | 0 |
meth |
Yuan_LinearLayer._tp_partition |
2 | 0 | 0 |
meth |
Conv_LinearALlreduce._tp_partition |
2 | 0 | 0 |
meth |
conv_LinearLayer._tp_partition |
2 | 0 | 0 |
meth |
LinearAllreduce.init |
4 | 0 | 0 |
meth |
LinearAllreduce.forward |
2 | 0 | 0 |
meth |
LinearAllreduce.gather_params |
2 | 0 | 0 |
meth |
LinearAllreduce._tp_partition |
2 | 0 | 0 |
meth |
LinearAllreduce.uneven_partition |
2 | 0 | 0 |
attr |
LinearAllreduce.weight |
1 | 0 | 0 |
attr |
LinearAllreduce.bias |
1 | 0 | 0 |
meth |
SubParamLinearLayer.init |
6 | 0 | 0 |
meth |
SubParamLinearLayer.forward |
2 | 0 | 0 |
meth |
SubParamLinearLayer.gather_params |
2 | 0 | 0 |
meth |
SubParamLinearLayer._tp_partition |
2 | 0 | 0 |
attr |
SubParamLinearLayer.weight |
1 | 0 | 0 |
attr |
SubParamLinearLayer.bias |
1 | 0 | 0 |
attr |
SubParamLinearLayer.shape |
1 | 0 | 0 |
attr |
SubParamLinearLayer.partition_dim |
1 | 0 | 0 |
meth |
LmHeadLinearAllreduce.init |
4 | 0 | 0 |
meth |
LmHeadLinearAllreduce.forward |
2 | 0 | 0 |
attr |
LmHeadLinearAllreduce.name |
1 | 0 | 0 |
meth |
SubParamLinearAllreduce.init |
6 | 0 | 0 |
meth |
SubParamLinearAllreduce.forward |
2 | 0 | 0 |
meth |
SubParamLinearAllreduce.gather_params |
2 | 0 | 0 |
meth |
SubParamLinearAllreduce._tp_partition |
2 | 0 | 0 |
attr |
SubParamLinearAllreduce.weight |
1 | 0 | 0 |
attr |
SubParamLinearAllreduce.bias |
1 | 0 | 0 |
attr |
SubParamLinearAllreduce.shape |
1 | 0 | 0 |
attr |
SubParamLinearAllreduce.partition_dim |
1 | 0 | 0 |
meth |
Yuan_LinearAllreduce._tp_partition |
2 | 0 | 0 |
meth |
fused_LinearLayer.init |
5 | 0 | 0 |
meth |
fused_LinearLayer._tp_partition |
2 | 0 | 0 |
attr |
fused_LinearLayer.fused_module |
1 | 0 | 0 |
deepspeed.module_inject.load_checkpoint (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
load_model_with_checkpoint |
9 | 0 | 0 |
deepspeed.module_inject.module_quantize (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
quantize_module |
4 | 0 | 0 |
func |
quantize_transformer_layer |
5 | 0 | 0 |
deepspeed.module_inject.policy (61 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
maybe_get_lora |
2 | 0 | 0 |
meth |
TransformerPolicy.init |
11 | 0 | 0 |
meth |
TransformerPolicy.attention |
1 | 0 | 0 |
meth |
TransformerPolicy.get_hidden_heads |
1 | 0 | 0 |
meth |
TransformerPolicy.mlp |
1 | 0 | 0 |
meth |
TransformerPolicy.layernorm |
1 | 0 | 0 |
attr |
TransformerPolicy.cuda_graph_supported |
1 | 0 | 0 |
attr |
TransformerPolicy.inference |
1 | 0 | 0 |
attr |
TransformerPolicy.linear_layer |
1 | 0 | 0 |
attr |
TransformerPolicy.scale_attention |
1 | 0 | 0 |
attr |
TransformerPolicy.is_megatron_v2 |
1 | 0 | 0 |
attr |
TransformerPolicy.use_mup |
1 | 0 | 0 |
attr |
TransformerPolicy.mlp_act_func_type |
1 | 0 | 0 |
attr |
TransformerPolicy.pre_attn_norm |
1 | 0 | 0 |
attr |
TransformerPolicy.use_load_prefix |
1 | 0 | 0 |
attr |
TransformerPolicy.split_qkv |
1 | 0 | 0 |
attr |
TransformerPolicy.norm_type |
1 | 0 | 0 |
func |
pack_lora_weights |
2 | 0 | 0 |
meth |
DSPolicy.init |
1 | 0 | 0 |
meth |
DSPolicy.attention |
1 | 0 | 0 |
attr |
DSPolicy.cuda_graph_supported |
1 | 0 | 0 |
func |
transpose |
2 | 0 | 0 |
func |
maybe_copy_qkv |
8 | 0 | 0 |
func |
maybe_copy |
11 | 0 | 0 |
func |
maybe_copy_geglu |
7 | 0 | 0 |
deepspeed.module_inject.replace_module (40 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
skip_level_0_prefix |
3 | 0 | 0 |
func |
revert_transformer_layer |
5 | 0 | 0 |
meth |
GroupQuantizer.init |
5 | 0 | 0 |
meth |
GroupQuantizer.quantize |
5 | 0 | 0 |
attr |
GroupQuantizer.group_size |
1 | 0 | 0 |
attr |
GroupQuantizer.num_bits |
1 | 0 | 0 |
attr |
GroupQuantizer.q_int8 |
1 | 0 | 0 |
attr |
GroupQuantizer.num_groups |
1 | 0 | 0 |
func |
get_transformer_name |
2 | 0 | 0 |
func |
replace_module |
6 | 0 | 0 |
func |
generic_injection |
4 | 0 | 0 |
func |
replace_transformer_layer |
6 | 0 | 0 |
deepspeed.module_inject.tp_shard (20 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_num_attention_heads |
1 | 0 | 0 |
func |
set_num_attention_heads |
2 | 0 | 0 |
func |
get_shard_size |
5 | 0 | 0 |
func |
set_num_kv_heads |
2 | 0 | 0 |
func |
set_tp_grain_size |
2 | 0 | 0 |
func |
get_n_embd |
1 | 0 | 0 |
func |
set_n_embd |
2 | 0 | 0 |
func |
get_shard_size_list |
4 | 0 | 0 |
func |
get_num_kv_heads |
1 | 0 | 0 |
deepspeed.module_inject.utils (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
policy_to_ds_container |
2 | 0 | 0 |
deepspeed.moe.experts (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
Experts.deepspeed_experts |
1 | 0 | 0 |
attr |
Experts.num_local_experts |
1 | 0 | 0 |
deepspeed.moe.layer (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
MoE.use_residual |
1 | 0 | 0 |
attr |
MoE.enable_expert_tensor_parallelism |
1 | 0 | 0 |
attr |
MoE.ep_size |
1 | 0 | 0 |
attr |
MoE.expert_group_name |
1 | 0 | 0 |
attr |
MoE.num_experts |
1 | 0 | 0 |
attr |
MoE.num_local_experts |
1 | 0 | 0 |
attr |
MoE.deepspeed_moe |
1 | 0 | 0 |
attr |
MoE.mlp |
1 | 0 | 0 |
attr |
MoE.coefficient |
1 | 0 | 0 |
deepspeed.moe.mappings (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
gather_tokens |
3 | 0 | 0 |
func |
drop_tokens |
3 | 0 | 0 |
deepspeed.moe.sharded_moe (37 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
multiplicative_jitter |
4 | 1 | 0 |
meth |
TopKGate._set_ep_group |
2 | 0 | 0 |
attr |
TopKGate.ep_group |
1 | 0 | 0 |
attr |
TopKGate.k |
1 | 0 | 0 |
attr |
TopKGate.capacity_factor |
1 | 0 | 0 |
attr |
TopKGate.eval_capacity_factor |
1 | 0 | 0 |
attr |
TopKGate.min_capacity |
1 | 0 | 0 |
attr |
TopKGate.noisy_gate_policy |
1 | 0 | 0 |
attr |
TopKGate.timers |
1 | 0 | 0 |
attr |
TopKGate.wall_clock_breakdown |
1 | 0 | 0 |
attr |
TopKGate.gate_time |
1 | 0 | 0 |
attr |
TopKGate.drop_tokens |
1 | 0 | 0 |
attr |
TopKGate.use_rts |
1 | 0 | 0 |
attr |
TopKGate.top2_2nd_expert_sampling |
1 | 0 | 0 |
func |
einsum |
4 | 0 | 0 |
meth |
MOELayer.init |
7 | 5 | 0 |
meth |
MOELayer._set_ep_group |
2 | 0 | 0 |
meth |
MOELayer.forward |
3 | 3 | 1 |
attr |
MOELayer.gate |
1 | 0 | 0 |
attr |
MOELayer.experts |
1 | 0 | 0 |
attr |
MOELayer.ep_group |
1 | 0 | 0 |
attr |
MOELayer.ep_size |
1 | 0 | 0 |
attr |
MOELayer.ep_group_name |
1 | 0 | 0 |
attr |
MOELayer.num_local_experts |
1 | 0 | 0 |
attr |
MOELayer.time_falltoall |
1 | 0 | 0 |
attr |
MOELayer.time_salltoall |
1 | 0 | 0 |
attr |
MOELayer.time_moe |
1 | 0 | 0 |
attr |
MOELayer.timers |
1 | 0 | 0 |
attr |
MOELayer.wall_clock_breakdown |
1 | 0 | 0 |
attr |
MOELayer.use_tutel |
1 | 0 | 0 |
deepspeed.moe.utils (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
is_moe_param_group |
2 | 0 | 0 |
func |
configure_moe_param_groups |
2 | 1 | 0 |
deepspeed.monitor.comet (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CometMonitor.init |
2 | 1 | 0 |
attr |
CometMonitor.enabled |
1 | 0 | 0 |
meth |
EventsLogScheduler.init |
2 | 1 | 0 |
deepspeed.monitor.config (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_monitor_config |
2 | 0 | 0 |
meth |
DeepSpeedMonitorConfig.check_enabled |
1 | 0 | 0 |
deepspeed.monitor.csv_monitor (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
csvMonitor.init |
2 | 0 | 0 |
meth |
csvMonitor.setup_log_dir |
2 | 0 | 0 |
meth |
csvMonitor.write_events |
2 | 0 | 0 |
attr |
csvMonitor.filenames |
1 | 0 | 0 |
attr |
csvMonitor.enabled |
1 | 0 | 0 |
attr |
csvMonitor.output_path |
1 | 0 | 0 |
attr |
csvMonitor.job_name |
1 | 0 | 0 |
attr |
csvMonitor.log_dir |
1 | 0 | 0 |
deepspeed.monitor.monitor (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MonitorMaster.init |
2 | 0 | 0 |
meth |
MonitorMaster.write_events |
2 | 0 | 0 |
attr |
MonitorMaster.tb_monitor |
1 | 0 | 0 |
attr |
MonitorMaster.wandb_monitor |
1 | 0 | 0 |
attr |
MonitorMaster.csv_monitor |
1 | 0 | 0 |
attr |
MonitorMaster.comet_monitor |
1 | 0 | 0 |
attr |
MonitorMaster.enabled |
1 | 0 | 0 |
meth |
Monitor.init |
2 | 0 | 0 |
meth |
Monitor.write_events |
2 | 0 | 0 |
attr |
Monitor.monitor_config |
1 | 0 | 0 |
deepspeed.monitor.tensorboard (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TensorBoardMonitor.init |
2 | 0 | 0 |
meth |
TensorBoardMonitor.get_summary_writer |
2 | 0 | 0 |
meth |
TensorBoardMonitor.write_events |
3 | 0 | 0 |
meth |
TensorBoardMonitor.flush |
1 | 0 | 0 |
attr |
TensorBoardMonitor.summary_writer |
1 | 0 | 0 |
attr |
TensorBoardMonitor.enabled |
1 | 0 | 0 |
attr |
TensorBoardMonitor.output_path |
1 | 0 | 0 |
attr |
TensorBoardMonitor.job_name |
1 | 0 | 0 |
deepspeed.monitor.utils (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
check_comet_availability |
1 | 0 | 0 |
func |
check_wandb_availability |
1 | 0 | 0 |
func |
check_tb_availability |
1 | 0 | 0 |
deepspeed.monitor.wandb (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
WandbMonitor.init |
2 | 0 | 0 |
meth |
WandbMonitor.log |
4 | 0 | 0 |
meth |
WandbMonitor.write_events |
2 | 0 | 0 |
attr |
WandbMonitor.enabled |
1 | 0 | 0 |
attr |
WandbMonitor.group |
1 | 0 | 0 |
attr |
WandbMonitor.team |
1 | 0 | 0 |
attr |
WandbMonitor.project |
1 | 0 | 0 |
deepspeed.nebula.config (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedNebulaConfig.init |
2 | 0 | 0 |
meth |
DeepSpeedNebulaConfig._initialize |
2 | 0 | 0 |
attr |
DeepSpeedNebulaConfig.enabled |
1 | 0 | 0 |
attr |
DeepSpeedNebulaConfig.persistent_storage_path |
1 | 0 | 0 |
attr |
DeepSpeedNebulaConfig.persistent_time_interval |
1 | 0 | 0 |
attr |
DeepSpeedNebulaConfig.num_of_version_in_retention |
1 | 0 | 0 |
attr |
DeepSpeedNebulaConfig.enable_nebula_load |
1 | 0 | 0 |
deepspeed.nvme.ds_aio_args (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
validate_args |
2 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
func |
get_validated_args |
1 | 0 | 0 |
func |
refine_args |
2 | 0 | 0 |
deepspeed.nvme.ds_aio_basic (18 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AIOBasic_Engine.init |
4 | 0 | 0 |
meth |
AIOBasic_Engine.fini |
1 | 0 | 0 |
meth |
AIOBasic_Engine.read |
4 | 0 | 0 |
meth |
AIOBasic_Engine.write |
4 | 0 | 0 |
meth |
AIOBasic_Engine._create_context |
4 | 0 | 0 |
attr |
AIOBasic_Engine.ctxt |
1 | 0 | 0 |
deepspeed.nvme.ds_aio_handle (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AIOHandle_Engine.init |
4 | 0 | 0 |
meth |
AIOHandle_Engine.fini |
1 | 0 | 0 |
meth |
AIOHandle_Engine.read |
4 | 0 | 0 |
meth |
AIOHandle_Engine.write |
4 | 0 | 0 |
meth |
AIOHandle_Engine._create_files |
4 | 0 | 0 |
meth |
AIOHandle_Engine._create_context |
4 | 0 | 0 |
attr |
AIOHandle_Engine.ctxt |
1 | 0 | 0 |
deepspeed.nvme.ds_aio_job (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
run_job |
3 | 0 | 0 |
meth |
Job.init |
4 | 0 | 0 |
meth |
Job.cmd |
1 | 0 | 0 |
meth |
Job.get_stdout |
1 | 0 | 0 |
meth |
Job.get_stderr |
1 | 0 | 0 |
meth |
Job.get_cwd |
1 | 0 | 0 |
meth |
Job.open_output_file |
1 | 0 | 0 |
meth |
Job.close_output_file |
1 | 0 | 0 |
attr |
Job.cmd_line |
1 | 0 | 0 |
attr |
Job.output_file |
1 | 0 | 0 |
attr |
Job.work_dir |
1 | 0 | 0 |
attr |
Job.output_fd |
1 | 0 | 0 |
deepspeed.nvme.io_engine (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_schedule |
3 | 0 | 0 |
func |
prepare_write |
2 | 0 | 0 |
func |
read_operation |
2 | 0 | 0 |
func |
prepare_read |
2 | 0 | 0 |
func |
io_engine_tasklet |
2 | 0 | 0 |
func |
prepare_operation |
4 | 0 | 0 |
func |
io_engine_multiprocessing |
3 | 0 | 0 |
func |
post_operation |
2 | 0 | 0 |
func |
write_operation |
2 | 0 | 0 |
deepspeed.nvme.parse_nvme_stats (20 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_metric |
3 | 0 | 0 |
func |
get_thread_count |
2 | 0 | 0 |
func |
get_results |
3 | 0 | 0 |
func |
extract_value |
3 | 0 | 0 |
func |
get_file_key |
2 | 0 | 0 |
func |
validate_args |
2 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
func |
get_sorted_results |
3 | 0 | 0 |
func |
main |
1 | 0 | 0 |
deepspeed.nvme.perf_generate_param (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
generate_aio_param |
3 | 0 | 0 |
func |
generate_main |
2 | 0 | 0 |
func |
validate_args |
2 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
func |
convert_to_param |
2 | 0 | 0 |
func |
main |
1 | 0 | 0 |
deepspeed.nvme.perf_run_sweep (53 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
run_read_sweep |
5 | 0 | 0 |
func |
get_log_file |
3 | 0 | 0 |
func |
run_write_sweep |
5 | 0 | 0 |
func |
sweep_main |
2 | 0 | 0 |
func |
create_perf_jobs |
4 | 0 | 0 |
func |
script_path |
1 | 0 | 0 |
func |
parse_sweep_arguments |
1 | 0 | 0 |
meth |
SweepConfig.init |
2 | 0 | 0 |
attr |
SweepConfig.folder_to_device_mapping |
1 | 0 | 0 |
attr |
SweepConfig.search_space |
1 | 0 | 0 |
attr |
SweepConfig.read |
1 | 0 | 0 |
attr |
SweepConfig.write |
1 | 0 | 0 |
attr |
SweepConfig.flush_cache |
1 | 0 | 0 |
attr |
SweepConfig.log_dir |
1 | 0 | 0 |
attr |
SweepConfig.verbose |
1 | 0 | 0 |
attr |
SweepConfig.other_options |
1 | 0 | 0 |
func |
gds_io_setup |
1 | 0 | 0 |
func |
get_ftd_map |
2 | 0 | 0 |
func |
validate_arguments |
2 | 0 | 0 |
func |
get_sweep_config_dict |
2 | 0 | 0 |
func |
get_sweep_cmd_lines |
2 | 0 | 0 |
func |
async_io_setup |
1 | 0 | 0 |
func |
main |
1 | 0 | 0 |
func |
launch_sweep |
5 | 0 | 0 |
func |
remove_folder |
2 | 0 | 0 |
func |
dump_cmd_lines |
2 | 0 | 0 |
func |
create_cmd_tags |
2 | 0 | 0 |
deepspeed.nvme.test_ds_aio (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
ds_io_main |
1 | 0 | 0 |
deepspeed.nvme.test_ds_aio_utils (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
task_barrier |
3 | 0 | 0 |
func |
create_filename |
5 | 0 | 0 |
func |
create_page_locked_tensor |
4 | 0 | 0 |
func |
report_results |
4 | 0 | 0 |
func |
create_file |
3 | 0 | 0 |
func |
task_log |
4 | 0 | 0 |
func |
refine_integer_value |
2 | 0 | 0 |
func |
get_block_size_and_count |
2 | 0 | 0 |
deepspeed.nvme.torch_fastio_engine (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Torch_FastIO_Engine.init |
4 | 0 | 0 |
meth |
Torch_FastIO_Engine.fini |
1 | 0 | 0 |
meth |
Torch_FastIO_Engine.read |
3 | 0 | 0 |
meth |
Torch_FastIO_Engine.write |
3 | 0 | 0 |
meth |
Torch_FastIO_Engine._create_context |
4 | 0 | 0 |
attr |
Torch_FastIO_Engine.ctxt |
1 | 0 | 0 |
attr |
Torch_FastIO_Engine.zipfile_serialization |
1 | 0 | 0 |
deepspeed.nvme.torch_io (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TorchIO_Engine.init |
4 | 0 | 0 |
meth |
TorchIO_Engine.fini |
1 | 0 | 0 |
meth |
TorchIO_Engine.read |
3 | 0 | 0 |
meth |
TorchIO_Engine.write |
3 | 0 | 0 |
meth |
TorchIO_Engine._create_context |
4 | 0 | 0 |
attr |
TorchIO_Engine.ctxt |
1 | 0 | 0 |
attr |
TorchIO_Engine.zipfile_serialization |
1 | 0 | 0 |
deepspeed.ops.adagrad.cpu_adagrad (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedCPUAdagrad.init |
7 | 0 | 0 |
meth |
DeepSpeedCPUAdagrad.del |
1 | 0 | 0 |
meth |
DeepSpeedCPUAdagrad.setstate |
2 | 0 | 0 |
meth |
DeepSpeedCPUAdagrad.step |
2 | 0 | 0 |
attr |
DeepSpeedCPUAdagrad.opt_id |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdagrad.fp32_optimizer_states |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdagrad.ds_opt_adagrad |
1 | 0 | 0 |
deepspeed.ops.adam.cpu_adam (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedCPUAdam.init |
10 | 0 | 0 |
meth |
DeepSpeedCPUAdam.del |
1 | 0 | 0 |
meth |
DeepSpeedCPUAdam.setstate |
2 | 0 | 0 |
meth |
DeepSpeedCPUAdam.step |
2 | 0 | 0 |
meth |
DeepSpeedCPUAdam.step_subgroup |
3 | 1 | 0 |
meth |
DeepSpeedCPUAdam.rollback_subgroup |
3 | 1 | 0 |
attr |
DeepSpeedCPUAdam.cpu_vendor |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdam.opt_id |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdam.adam_w_mode |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdam.fp32_optimizer_states |
1 | 0 | 0 |
attr |
DeepSpeedCPUAdam.ds_opt_adam |
1 | 0 | 0 |
deepspeed.ops.adam.fused_adam (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
multi_tensor_applier |
1 | 0 | 0 |
meth |
FusedAdam.init |
10 | 0 | 0 |
meth |
FusedAdam.zero_grad |
1 | 0 | 0 |
meth |
FusedAdam.step |
7 | 0 | 0 |
attr |
FusedAdam.adam_w_mode |
1 | 0 | 0 |
attr |
FusedAdam.set_grad_none |
1 | 0 | 0 |
attr |
FusedAdam.multi_tensor_adam |
1 | 0 | 0 |
deepspeed.ops.adam.multi_tensor_apply (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MultiTensorApply.init |
2 | 0 | 0 |
meth |
MultiTensorApply.call |
5 | 0 | 0 |
attr |
MultiTensorApply.chunk_size |
1 | 0 | 0 |
deepspeed.ops.adam.zenflow_cpu_adam (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZenFlowCPUAdam.init |
4 | 0 | 0 |
meth |
ZenFlowCPUAdam._sequential_step |
3 | 0 | 0 |
meth |
ZenFlowCPUAdam._parallel_step |
5 | 0 | 0 |
attr |
ZenFlowCPUAdam.overlap_step |
1 | 0 | 0 |
attr |
ZenFlowCPUAdam.step |
1 | 0 | 0 |
deepspeed.ops.adam.zenflow_torch_adam (39 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZenFlowSelectiveAdamW_stage3.init |
5 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3.temp_copy_param |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3.clear_selected_mv |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3._step_without_offload |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3.copy_mv_from_cpu |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3.copy_mv_to_cpu |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3.group_step |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW_stage3._step_with_offload |
1 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW_stage3.offload |
1 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW_stage3.step |
1 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW_stage3.bucket_size |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.init |
5 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.temp_copy_param |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.copy_mv_from_cpu |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.copy_mv_to_cpu |
2 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.clear_selected_mv |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW._step_without_offload |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW._step_with_offload |
1 | 0 | 0 |
meth |
ZenFlowSelectiveAdamW.group_step |
2 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW.offload |
1 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW.step |
1 | 0 | 0 |
attr |
ZenFlowSelectiveAdamW.bucket_size |
1 | 0 | 0 |
func |
adamw |
21 | 20 | 0 |
deepspeed.ops.deepspeed4science.evoformer_attn (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
EvoformerFusedAttention.forward |
7 | 0 | 0 |
meth |
EvoformerFusedAttention.backward |
3 | 0 | 0 |
func |
DS4Sci_EvoformerAttention |
5 | 0 | 0 |
func |
attention_bwd |
11 | 0 | 0 |
deepspeed.ops.fp_quantizer.fp8_gemm (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
matmul_fp8_fallback |
6 | 0 | 0 |
func |
matmul_fp8 |
6 | 0 | 0 |
deepspeed.ops.fp_quantizer.fp8_gemm_triton (33 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
matmul_kernel_fp8_fp16 |
19 | 5 | 0 |
func |
matmul_fp8_triton |
5 | 0 | 0 |
func |
matmul_kernel_fp8_bf16 |
19 | 5 | 0 |
deepspeed.ops.fp_quantizer.quantize (39 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Quantizer.init |
2 | 1 | 0 |
meth |
Quantizer.quantize |
6 | 1 | 0 |
meth |
Quantizer.dequantize |
6 | 1 | 0 |
attr |
Quantizer.group_size |
1 | 0 | 0 |
meth |
FP_Quantize.init |
2 | 1 | 0 |
meth |
FP_Quantize.quantize |
6 | 1 | 0 |
meth |
FP_Quantize.to |
3 | 0 | 0 |
meth |
FP_Quantize.get_scales |
1 | 0 | 0 |
meth |
FP_Quantize.dequantize |
6 | 1 | 0 |
meth |
FP_Quantize.selective_dequantize |
7 | 1 | 0 |
attr |
FP_Quantize.cuda_impl |
1 | 0 | 0 |
attr |
FP_Quantize.q_config |
1 | 0 | 0 |
attr |
FP_Quantize.orig_dtype |
1 | 0 | 0 |
attr |
FP_Quantize.num_groups |
1 | 0 | 0 |
attr |
FP_Quantize.input_q |
1 | 0 | 0 |
attr |
FP_Quantize.scale |
1 | 0 | 0 |
deepspeed.ops.lamb.fused_lamb (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedLamb.init |
12 | 0 | 0 |
meth |
FusedLamb.step |
6 | 0 | 0 |
meth |
FusedLamb.get_lamb_coeffs |
1 | 0 | 0 |
attr |
FusedLamb.fused_lamb_cuda |
1 | 0 | 0 |
attr |
FusedLamb.eps_mode |
1 | 0 | 0 |
attr |
FusedLamb.lamb_coeffs |
1 | 0 | 0 |
deepspeed.ops.lion.cpu_lion (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedCPULion.init |
6 | 0 | 0 |
meth |
DeepSpeedCPULion.del |
1 | 0 | 0 |
meth |
DeepSpeedCPULion.setstate |
2 | 0 | 0 |
meth |
DeepSpeedCPULion.step |
2 | 0 | 0 |
attr |
DeepSpeedCPULion.cpu_vendor |
1 | 0 | 0 |
attr |
DeepSpeedCPULion.opt_id |
1 | 0 | 0 |
attr |
DeepSpeedCPULion.fp32_optimizer_states |
1 | 0 | 0 |
attr |
DeepSpeedCPULion.ds_opt_lion |
1 | 0 | 0 |
deepspeed.ops.lion.fused_lion (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedLion.init |
6 | 0 | 0 |
meth |
FusedLion.zero_grad |
1 | 0 | 0 |
meth |
FusedLion.step |
7 | 0 | 0 |
attr |
FusedLion.set_grad_none |
1 | 0 | 0 |
attr |
FusedLion.multi_tensor_lion |
1 | 0 | 0 |
attr |
multi_tensor_applier |
1 | 0 | 0 |
deepspeed.ops.lion.multi_tensor_apply (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MultiTensorApply.init |
2 | 0 | 0 |
meth |
MultiTensorApply.call |
5 | 0 | 0 |
attr |
MultiTensorApply.chunk_size |
1 | 0 | 0 |
deepspeed.ops.op_builder (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
module |
1 | 0 | 0 |
func |
builder_closure |
2 | 0 | 0 |
deepspeed.ops.op_builder.all_ops (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
op_builder_module |
1 | 0 | 0 |
attr |
module |
1 | 0 | 0 |
attr |
op_builder_dir |
1 | 0 | 0 |
attr |
builder |
1 | 0 | 0 |
deepspeed.ops.op_builder.async_io (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AsyncIOBuilder.init |
1 | 0 | 0 |
meth |
AsyncIOBuilder.absolute_name |
1 | 0 | 0 |
meth |
AsyncIOBuilder.lib_sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.include_paths |
1 | 0 | 0 |
meth |
AsyncIOBuilder.cxx_args |
1 | 0 | 0 |
meth |
AsyncIOBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
AsyncIOBuilder.check_for_libaio_pkg |
1 | 0 | 0 |
meth |
AsyncIOBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.builder (69 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
installed_cuda_version |
2 | 0 | 0 |
func |
assert_no_cuda_mismatch |
2 | 0 | 0 |
attr |
TORCH_MINOR |
1 | 0 | 0 |
func |
get_default_compute_capabilities |
1 | 0 | 0 |
meth |
OpBuilder.init |
2 | 0 | 0 |
meth |
OpBuilder.absolute_name |
1 | 0 | 0 |
meth |
OpBuilder.sources |
1 | 0 | 0 |
meth |
OpBuilder.hipify_extension |
1 | 0 | 0 |
meth |
OpBuilder.sycl_extension |
1 | 0 | 0 |
meth |
OpBuilder.validate_torch_version |
2 | 0 | 0 |
meth |
OpBuilder.validate_torch_op_version |
2 | 0 | 0 |
meth |
OpBuilder.is_rocm_pytorch |
1 | 0 | 0 |
meth |
OpBuilder.is_sycl_enabled |
1 | 0 | 0 |
meth |
OpBuilder.installed_rocm_version |
1 | 0 | 0 |
meth |
OpBuilder.get_rocm_gpu_arch |
1 | 0 | 0 |
meth |
OpBuilder.get_rocm_wavefront_size |
1 | 0 | 0 |
meth |
OpBuilder.include_paths |
1 | 0 | 0 |
meth |
OpBuilder.nvcc_args |
1 | 0 | 0 |
meth |
OpBuilder.cxx_args |
1 | 0 | 0 |
meth |
OpBuilder.is_compatible |
2 | 0 | 0 |
meth |
OpBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
OpBuilder.has_function |
5 | 0 | 0 |
meth |
OpBuilder.strip_empty_entries |
2 | 0 | 0 |
meth |
OpBuilder.cpu_arch |
1 | 0 | 0 |
meth |
OpBuilder.get_cuda_compile_flag |
1 | 0 | 0 |
meth |
OpBuilder._backup_cpuinfo |
1 | 0 | 0 |
meth |
OpBuilder.simd_width |
1 | 0 | 0 |
meth |
OpBuilder.command_exists |
2 | 0 | 0 |
meth |
OpBuilder.warning |
2 | 0 | 0 |
meth |
OpBuilder.deepspeed_src_path |
2 | 0 | 0 |
meth |
OpBuilder.builder |
1 | 0 | 0 |
meth |
OpBuilder.load |
2 | 0 | 0 |
meth |
OpBuilder.jit_load |
2 | 0 | 0 |
attr |
OpBuilder.name |
1 | 0 | 0 |
attr |
OpBuilder.jit_mode |
1 | 0 | 0 |
attr |
OpBuilder.build_for_cpu |
1 | 0 | 0 |
attr |
OpBuilder.enable_bf16 |
1 | 0 | 0 |
attr |
OpBuilder.error_log |
1 | 0 | 0 |
attr |
TORCH_MAJOR |
1 | 0 | 0 |
meth |
TorchCPUOpBuilder.get_cuda_lib64_path |
1 | 0 | 0 |
meth |
TorchCPUOpBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
TorchCPUOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
CUDAOpBuilder.compute_capability_args |
2 | 0 | 0 |
meth |
CUDAOpBuilder.filter_ccs |
2 | 1 | 0 |
meth |
CUDAOpBuilder.version_dependent_macros |
1 | 0 | 0 |
meth |
CUDAOpBuilder.is_compatible |
2 | 0 | 0 |
meth |
CUDAOpBuilder.builder |
1 | 0 | 0 |
meth |
CUDAOpBuilder.hipify_extension |
1 | 0 | 0 |
meth |
CUDAOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
CUDAOpBuilder.nvcc_args |
1 | 0 | 0 |
meth |
CUDAOpBuilder.libraries_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu.async_io (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AsyncIOBuilder.init |
1 | 0 | 0 |
meth |
AsyncIOBuilder.absolute_name |
1 | 0 | 0 |
meth |
AsyncIOBuilder.lib_sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.include_paths |
1 | 0 | 0 |
meth |
AsyncIOBuilder.cxx_args |
1 | 0 | 0 |
meth |
AsyncIOBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
AsyncIOBuilder.check_for_libaio_pkg |
1 | 0 | 0 |
meth |
AsyncIOBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.cpu.builder (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUOpBuilder.builder |
1 | 0 | 0 |
meth |
CPUOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
CPUOpBuilder.libraries_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu.comm (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ShareMemCommBuilder.init |
2 | 0 | 0 |
meth |
ShareMemCommBuilder.absolute_name |
1 | 0 | 0 |
meth |
ShareMemCommBuilder.sources |
1 | 0 | 0 |
meth |
ShareMemCommBuilder.include_paths |
1 | 0 | 0 |
meth |
ShareMemCommBuilder.cxx_args |
1 | 0 | 0 |
meth |
ShareMemCommBuilder.is_compatible |
2 | 0 | 0 |
meth |
CCLCommBuilder.init |
2 | 0 | 0 |
meth |
CCLCommBuilder.absolute_name |
1 | 0 | 0 |
meth |
CCLCommBuilder.sources |
1 | 0 | 0 |
meth |
CCLCommBuilder.include_paths |
1 | 0 | 0 |
meth |
CCLCommBuilder.cxx_args |
1 | 0 | 0 |
meth |
CCLCommBuilder.is_compatible |
2 | 0 | 0 |
meth |
CCLCommBuilder.extra_ldflags |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu.fused_adam (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu.no_impl (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu_adagrad (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdagradBuilder.init |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.cpu_lion (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPULionBuilder.init |
1 | 0 | 0 |
meth |
CPULionBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPULionBuilder.sources |
1 | 0 | 0 |
meth |
CPULionBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPULionBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.dc (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepCompileBuilder.init |
1 | 0 | 0 |
meth |
DeepCompileBuilder.absolute_name |
1 | 0 | 0 |
meth |
DeepCompileBuilder.sources |
1 | 0 | 0 |
meth |
DeepCompileBuilder.libraries_args |
1 | 0 | 0 |
meth |
DeepCompileBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.evoformer_attn (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
EvoformerAttnBuilder.init |
2 | 0 | 0 |
meth |
EvoformerAttnBuilder.absolute_name |
1 | 0 | 0 |
meth |
EvoformerAttnBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
EvoformerAttnBuilder.sources |
1 | 0 | 0 |
meth |
EvoformerAttnBuilder.nvcc_args |
1 | 0 | 0 |
meth |
EvoformerAttnBuilder.filter_ccs |
2 | 0 | 0 |
meth |
EvoformerAttnBuilder.is_compatible |
2 | 0 | 0 |
meth |
EvoformerAttnBuilder.include_paths |
1 | 0 | 0 |
attr |
EvoformerAttnBuilder.cutlass_path |
1 | 0 | 0 |
deepspeed.ops.op_builder.fp_quantizer (13 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FPQuantizerBuilder.init |
2 | 0 | 0 |
meth |
FPQuantizerBuilder.absolute_name |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.is_compatible |
2 | 0 | 0 |
meth |
FPQuantizerBuilder.filter_ccs |
2 | 0 | 0 |
meth |
FPQuantizerBuilder.sources |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.include_paths |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.get_default_quant_dtype |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.get_quant_range |
2 | 0 | 0 |
deepspeed.ops.op_builder.fused_adam (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.cxx_args |
1 | 0 | 0 |
meth |
FusedAdamBuilder.nvcc_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.fused_lamb (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedLambBuilder.init |
1 | 0 | 0 |
meth |
FusedLambBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedLambBuilder.sources |
1 | 0 | 0 |
meth |
FusedLambBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedLambBuilder.cxx_args |
1 | 0 | 0 |
meth |
FusedLambBuilder.nvcc_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.fused_lion (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedLionBuilder.init |
1 | 0 | 0 |
meth |
FusedLionBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedLionBuilder.sources |
1 | 0 | 0 |
meth |
FusedLionBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedLionBuilder.cxx_args |
1 | 0 | 0 |
meth |
FusedLionBuilder.nvcc_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.gds (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GDSBuilder.init |
1 | 0 | 0 |
meth |
GDSBuilder.absolute_name |
1 | 0 | 0 |
meth |
GDSBuilder.lib_sources |
1 | 0 | 0 |
meth |
GDSBuilder.sources |
1 | 0 | 0 |
meth |
GDSBuilder.cxx_args |
1 | 0 | 0 |
meth |
GDSBuilder.include_paths |
1 | 0 | 0 |
meth |
GDSBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
GDSBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.hpu.builder (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUOpBuilder.builder |
1 | 0 | 0 |
meth |
CPUOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
CPUOpBuilder.libraries_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.hpu.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.hpu.fp_quantizer (34 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FPQuantizerBuilder.init |
2 | 0 | 0 |
meth |
FPQuantizerBuilder.absolute_name |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.sources |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.load |
2 | 0 | 0 |
meth |
FPQuantizerBuilder.get_default_quant_dtype |
1 | 0 | 0 |
meth |
FPQuantizerBuilder.get_quant_range |
2 | 0 | 0 |
meth |
FPQuantizer.selective_dequantize |
7 | 0 | 0 |
meth |
FPQuantizer.dequantize |
7 | 0 | 0 |
meth |
FPQuantizer.quantize |
8 | 0 | 0 |
meth |
FPQuantizer.get_scales |
3 | 0 | 0 |
deepspeed.ops.op_builder.hpu.fused_adam (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HPUFusedAdam.multi_tensor_adam |
13 | 0 | 0 |
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.load |
2 | 0 | 0 |
deepspeed.ops.op_builder.hpu.no_impl (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
deepspeed.ops.op_builder.hpu.transformer_inference (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceBuilder.init |
2 | 0 | 0 |
meth |
InferenceBuilder.absolute_name |
1 | 0 | 0 |
meth |
InferenceBuilder.sources |
1 | 0 | 0 |
meth |
InferenceBuilder.load |
2 | 0 | 0 |
deepspeed.ops.op_builder.inference_core_ops (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceCoreBuilder.init |
2 | 0 | 0 |
meth |
InferenceCoreBuilder.absolute_name |
1 | 0 | 0 |
meth |
InferenceCoreBuilder.is_compatible |
2 | 0 | 0 |
meth |
InferenceCoreBuilder.filter_ccs |
2 | 0 | 0 |
meth |
InferenceCoreBuilder.get_prefix |
1 | 0 | 0 |
meth |
InferenceCoreBuilder.sources |
1 | 0 | 0 |
meth |
InferenceCoreBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
InferenceCoreBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.inference_cutlass_builder (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceCutlassBuilder.init |
2 | 0 | 0 |
meth |
InferenceCutlassBuilder.absolute_name |
1 | 0 | 0 |
meth |
InferenceCutlassBuilder.is_compatible |
2 | 0 | 0 |
meth |
InferenceCutlassBuilder.filter_ccs |
2 | 0 | 0 |
meth |
InferenceCutlassBuilder.get_prefix |
1 | 0 | 0 |
meth |
InferenceCutlassBuilder.sources |
1 | 0 | 0 |
meth |
InferenceCutlassBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
InferenceCutlassBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.mlu.builder (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MLUOpBuilder.builder |
1 | 0 | 0 |
meth |
MLUOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
MLUOpBuilder.libraries_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.mlu.cpu_adagrad (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdagradBuilder.init |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.mlu.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.mlu.fused_adam (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MLUFusedAdam.multi_tensor_adam |
13 | 0 | 0 |
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.load |
2 | 0 | 0 |
deepspeed.ops.op_builder.mlu.no_impl (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
meth |
NotImplementedBuilder.cxx_args |
1 | 0 | 0 |
meth |
NotImplementedBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
NotImplementedBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.npu.async_io (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AsyncIOBuilder.init |
1 | 0 | 0 |
meth |
AsyncIOBuilder.absolute_name |
1 | 0 | 0 |
meth |
AsyncIOBuilder.sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.include_paths |
1 | 0 | 0 |
meth |
AsyncIOBuilder.cxx_args |
1 | 0 | 0 |
meth |
AsyncIOBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
AsyncIOBuilder.check_for_libaio_pkg |
1 | 0 | 0 |
meth |
AsyncIOBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.npu.builder (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NPUOpBuilder.init |
2 | 0 | 0 |
meth |
NPUOpBuilder.cann_defs |
1 | 0 | 0 |
meth |
NPUOpBuilder.installed_cann_path |
1 | 0 | 0 |
meth |
NPUOpBuilder.installed_cann_version |
2 | 0 | 0 |
meth |
NPUOpBuilder.include_paths |
1 | 0 | 0 |
meth |
NPUOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
NPUOpBuilder.extra_ldflags |
1 | 0 | 0 |
deepspeed.ops.op_builder.npu.cpu_adagrad (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdagradBuilder.init |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.npu.cpu_adam (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.npu.cpu_lion (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPULionBuilder.init |
1 | 0 | 0 |
meth |
CPULionBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPULionBuilder.sources |
1 | 0 | 0 |
meth |
CPULionBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.npu.fused_adam (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NPUFusedAdam.multi_tensor_adam |
13 | 0 | 0 |
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.load |
2 | 0 | 0 |
deepspeed.ops.op_builder.npu.inference (275 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceContext.reset_tokens |
2 | 0 | 0 |
meth |
InferenceContext.current_tokens |
1 | 0 | 0 |
meth |
InferenceContext.GetWorkSpace |
1 | 0 | 0 |
meth |
InferenceBuilder.init |
1 | 0 | 0 |
meth |
InferenceBuilder.absolute_name |
1 | 0 | 0 |
meth |
InferenceBuilder.sources |
1 | 0 | 0 |
meth |
InferenceBuilder.include_paths |
1 | 0 | 0 |
meth |
InferenceBuilder.load |
2 | 0 | 0 |
meth |
NPUInference.layer_norm |
5 | 0 | 0 |
meth |
NPUInference._qkv_gemm |
11 | 0 | 0 |
meth |
NPUInference.qkv_gemm_fp16 |
11 | 0 | 0 |
meth |
NPUInference.qkv_gemm_bf16 |
11 | 0 | 0 |
meth |
NPUInference.qkv_gemm_fp32 |
11 | 0 | 0 |
meth |
NPUInference._bias_add_transform_0213 |
12 | 0 | 0 |
meth |
NPUInference._softmax_context |
17 | 0 | 0 |
meth |
NPUInference.softmax_context_fp16 |
17 | 0 | 0 |
meth |
NPUInference.softmax_context_bf16 |
17 | 0 | 0 |
meth |
NPUInference.softmax_context_fp32 |
17 | 0 | 0 |
meth |
NPUInference._vector_matmul |
7 | 0 | 0 |
meth |
NPUInference.vector_matmul_fp16 |
7 | 0 | 0 |
meth |
NPUInference.vector_matmul_bf16 |
7 | 0 | 0 |
meth |
NPUInference.vector_matmul_fp32 |
7 | 0 | 0 |
meth |
NPUInference._mlp_gemm |
17 | 0 | 0 |
meth |
NPUInference.mlp_gemm_fp16 |
17 | 0 | 0 |
meth |
NPUInference.mlp_gemm_bf16 |
17 | 0 | 0 |
meth |
NPUInference.mlp_gemm_fp32 |
17 | 0 | 0 |
meth |
NPUInference._residual_add_bias |
10 | 0 | 0 |
meth |
NPUInference.residual_add_bias_fp16 |
10 | 0 | 0 |
meth |
NPUInference.residual_add_bias_bf16 |
10 | 0 | 0 |
meth |
NPUInference.residual_add_bias_fp32 |
10 | 0 | 0 |
deepspeed.ops.op_builder.npu.no_impl (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
meth |
NotImplementedBuilder.cxx_args |
1 | 0 | 0 |
meth |
NotImplementedBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
NotImplementedBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.quantizer (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
QuantizerBuilder.init |
2 | 0 | 0 |
meth |
QuantizerBuilder.absolute_name |
1 | 0 | 0 |
meth |
QuantizerBuilder.sources |
1 | 0 | 0 |
meth |
QuantizerBuilder.include_paths |
1 | 0 | 0 |
meth |
QuantizerBuilder.extra_ldflags |
1 | 0 | 0 |
deepspeed.ops.op_builder.ragged_ops (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RaggedOpsBuilder.init |
2 | 0 | 0 |
meth |
RaggedOpsBuilder.absolute_name |
1 | 0 | 0 |
meth |
RaggedOpsBuilder.is_compatible |
2 | 0 | 0 |
meth |
RaggedOpsBuilder.filter_ccs |
2 | 0 | 0 |
meth |
RaggedOpsBuilder.get_prefix |
1 | 0 | 0 |
meth |
RaggedOpsBuilder.sources |
1 | 0 | 0 |
meth |
RaggedOpsBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
RaggedOpsBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.ragged_utils (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RaggedUtilsBuilder.init |
2 | 0 | 0 |
meth |
RaggedUtilsBuilder.absolute_name |
1 | 0 | 0 |
meth |
RaggedUtilsBuilder.is_compatible |
2 | 0 | 0 |
meth |
RaggedUtilsBuilder.filter_ccs |
2 | 0 | 0 |
meth |
RaggedUtilsBuilder.get_prefix |
1 | 0 | 0 |
meth |
RaggedUtilsBuilder.sources |
1 | 0 | 0 |
meth |
RaggedUtilsBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
RaggedUtilsBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.random_ltd (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RandomLTDBuilder.init |
2 | 0 | 0 |
meth |
RandomLTDBuilder.absolute_name |
1 | 0 | 0 |
meth |
RandomLTDBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
RandomLTDBuilder.sources |
1 | 0 | 0 |
meth |
RandomLTDBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.sdaa.builder (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SDAAOpBuilder.builder |
1 | 0 | 0 |
meth |
SDAAOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
SDAAOpBuilder.libraries_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.sdaa.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.sdaa.fused_adam (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SDAAFusedAdam.multi_tensor_adam |
13 | 0 | 0 |
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.load |
2 | 0 | 0 |
deepspeed.ops.op_builder.sdaa.no_impl (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
meth |
NotImplementedBuilder.cxx_args |
1 | 0 | 0 |
meth |
NotImplementedBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
NotImplementedBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.sparse_attn (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SparseAttnBuilder.init |
1 | 0 | 0 |
meth |
SparseAttnBuilder.absolute_name |
1 | 0 | 0 |
meth |
SparseAttnBuilder.sources |
1 | 0 | 0 |
meth |
SparseAttnBuilder.cxx_args |
1 | 0 | 0 |
meth |
SparseAttnBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.spatial_inference (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SpatialInferenceBuilder.init |
2 | 0 | 0 |
meth |
SpatialInferenceBuilder.absolute_name |
1 | 0 | 0 |
meth |
SpatialInferenceBuilder.is_compatible |
2 | 0 | 0 |
meth |
SpatialInferenceBuilder.sources |
1 | 0 | 0 |
meth |
SpatialInferenceBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.stochastic_transformer (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
StochasticTransformerBuilder.init |
1 | 0 | 0 |
meth |
StochasticTransformerBuilder.absolute_name |
1 | 0 | 0 |
meth |
StochasticTransformerBuilder.nvcc_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.transformer (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TransformerBuilder.init |
2 | 0 | 0 |
meth |
TransformerBuilder.absolute_name |
1 | 0 | 0 |
meth |
TransformerBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
TransformerBuilder.sources |
1 | 0 | 0 |
meth |
TransformerBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.transformer_inference (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceBuilder.init |
2 | 0 | 0 |
meth |
InferenceBuilder.absolute_name |
1 | 0 | 0 |
meth |
InferenceBuilder.is_compatible |
2 | 0 | 0 |
meth |
InferenceBuilder.filter_ccs |
2 | 0 | 0 |
meth |
InferenceBuilder.sources |
1 | 0 | 0 |
meth |
InferenceBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
InferenceBuilder.include_paths |
1 | 0 | 0 |
meth |
InferenceBuilder.nvcc_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.utils (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
UtilsBuilder.init |
1 | 0 | 0 |
meth |
UtilsBuilder.absolute_name |
1 | 0 | 0 |
meth |
UtilsBuilder.sources |
1 | 0 | 0 |
deepspeed.ops.op_builder.xpu.async_io (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AsyncIOBuilder.init |
1 | 0 | 0 |
meth |
AsyncIOBuilder.absolute_name |
1 | 0 | 0 |
meth |
AsyncIOBuilder.sources |
1 | 0 | 0 |
meth |
AsyncIOBuilder.include_paths |
1 | 0 | 0 |
meth |
AsyncIOBuilder.cxx_args |
1 | 0 | 0 |
meth |
AsyncIOBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
AsyncIOBuilder.check_for_libaio_pkg |
1 | 0 | 0 |
meth |
AsyncIOBuilder.is_compatible |
2 | 0 | 0 |
deepspeed.ops.op_builder.xpu.builder (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SYCLOpBuilder.builder |
1 | 0 | 0 |
meth |
SYCLOpBuilder.version_dependent_macros |
1 | 0 | 0 |
meth |
SYCLOpBuilder._sycl_env_paths |
1 | 0 | 0 |
meth |
SYCLOpBuilder.cxx_args |
1 | 0 | 0 |
meth |
SYCLOpBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
SYCLOpBuilder.fixed_aotflags |
1 | 0 | 0 |
meth |
SYCLOpBuilder.load |
2 | 0 | 0 |
meth |
SYCLOpBuilder.jit_load |
2 | 0 | 0 |
deepspeed.ops.op_builder.xpu.cpu_adagrad (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdagradBuilder.init |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdagradBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.xpu.cpu_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CPUAdamBuilder.init |
1 | 0 | 0 |
meth |
CPUAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
CPUAdamBuilder.sources |
1 | 0 | 0 |
meth |
CPUAdamBuilder.libraries_args |
1 | 0 | 0 |
meth |
CPUAdamBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.xpu.flash_attn (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FlashAttentionBuilder.init |
2 | 0 | 0 |
meth |
FlashAttentionBuilder.absolute_name |
1 | 0 | 0 |
meth |
FlashAttentionBuilder.sources |
1 | 0 | 0 |
meth |
FlashAttentionBuilder.include_paths |
1 | 0 | 0 |
meth |
FlashAttentionBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
FlashAttentionBuilder.cxx_args |
1 | 0 | 0 |
meth |
FlashAttentionBuilder.load |
1 | 0 | 0 |
meth |
FlashAttentionBuilderObject.init |
1 | 0 | 0 |
meth |
FlashAttentionBuilderObject.flash_attn_func_v2 |
7 | 0 | 0 |
deepspeed.ops.op_builder.xpu.fused_adam (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FusedAdamBuilder.init |
1 | 0 | 0 |
meth |
FusedAdamBuilder.absolute_name |
1 | 0 | 0 |
meth |
FusedAdamBuilder.sources |
1 | 0 | 0 |
meth |
FusedAdamBuilder.include_paths |
1 | 0 | 0 |
meth |
FusedAdamBuilder.cxx_args |
1 | 0 | 0 |
deepspeed.ops.op_builder.xpu.no_impl (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NotImplementedBuilder.init |
2 | 0 | 0 |
meth |
NotImplementedBuilder.absolute_name |
1 | 0 | 0 |
meth |
NotImplementedBuilder.load |
2 | 0 | 0 |
meth |
NotImplementedBuilder.sources |
1 | 0 | 0 |
meth |
NotImplementedBuilder.cxx_args |
1 | 0 | 0 |
meth |
NotImplementedBuilder.extra_ldflags |
1 | 0 | 0 |
meth |
NotImplementedBuilder.include_paths |
1 | 0 | 0 |
deepspeed.ops.op_builder.xpu.packbits (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PackbitsBuilder.init |
1 | 0 | 0 |
meth |
PackbitsBuilder.absolute_name |
1 | 0 | 0 |
meth |
PackbitsBuilder.sources |
1 | 0 | 0 |
meth |
PackbitsBuilder.include_paths |
1 | 0 | 0 |
meth |
PackbitsBuilder.cxx_args |
1 | 0 | 0 |
deepspeed.ops.quantizer.quantizer (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
ds_quantizer |
6 | 0 | 0 |
deepspeed.ops.random_ltd.dropping_utils (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
bert_sample_tokens |
7 | 6 | 0 |
meth |
GatherTokens.forward |
5 | 3 | 0 |
meth |
GatherTokens.backward |
4 | 2 | 0 |
func |
gpt_sample_tokens |
7 | 6 | 0 |
meth |
ScatterTokens.forward |
6 | 4 | 0 |
meth |
ScatterTokens.backward |
3 | 1 | 0 |
deepspeed.ops.sparse_attention.bert_sparse_self_attention (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BertSparseSelfAttention.init |
3 | 0 | 0 |
meth |
BertSparseSelfAttention.transpose_for_scores |
2 | 0 | 0 |
meth |
BertSparseSelfAttention.forward |
3 | 0 | 0 |
attr |
BertSparseSelfAttention.num_attention_heads |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.attention_head_size |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.all_head_size |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.query |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.key |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.value |
1 | 0 | 0 |
attr |
BertSparseSelfAttention.sparse_self_attention |
1 | 0 | 0 |
deepspeed.ops.sparse_attention.matmul (33 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MatMul.make_lut |
3 | 0 | 0 |
meth |
MatMul.init |
7 | 0 | 0 |
meth |
MatMul._pad_shape |
3 | 0 | 0 |
meth |
MatMul.call |
3 | 0 | 0 |
meth |
MatMul._validate_inputs |
3 | 0 | 0 |
attr |
MatMul.lut_cache |
1 | 0 | 0 |
attr |
MatMul.trans_a |
1 | 0 | 0 |
attr |
MatMul.trans_b |
1 | 0 | 0 |
attr |
MatMul.mode |
1 | 0 | 0 |
attr |
MatMul.block |
1 | 0 | 0 |
attr |
MatMul.layout |
1 | 0 | 0 |
attr |
MatMul.spdims |
1 | 0 | 0 |
attr |
MatMul.bench |
1 | 0 | 0 |
attr |
MatMul.time_c |
1 | 0 | 0 |
attr |
MatMul.time_da |
1 | 0 | 0 |
attr |
MatMul.time_db |
1 | 0 | 0 |
attr |
MatMul.dense_inner_dim |
1 | 0 | 0 |
attr |
MatMul.dense_inner_size |
1 | 0 | 0 |
attr |
MatMul.sparse_shape |
1 | 0 | 0 |
deepspeed.ops.sparse_attention.softmax (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Softmax.sparse_softmax |
3 | 0 | 0 |
meth |
Softmax.make_lut |
2 | 0 | 0 |
meth |
Softmax.init |
4 | 0 | 0 |
meth |
Softmax.call |
8 | 0 | 0 |
attr |
Softmax.num_blocks |
1 | 0 | 0 |
attr |
Softmax.spdims |
1 | 0 | 0 |
attr |
Softmax.layout |
1 | 0 | 0 |
attr |
Softmax.block |
1 | 0 | 0 |
attr |
Softmax.bench |
1 | 0 | 0 |
attr |
Softmax.lut_cache |
1 | 0 | 0 |
func |
next_power_of_2 |
2 | 0 | 0 |
func |
num_warps |
2 | 0 | 0 |
deepspeed.ops.sparse_attention.sparse_attention_utils (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SparseAttentionUtils.extend_position_embedding |
3 | 0 | 0 |
meth |
SparseAttentionUtils.update_tokenizer_model_max_length |
3 | 0 | 0 |
meth |
SparseAttentionUtils.replace_model_self_attention_with_sparse_self_attention |
4 | 0 | 0 |
meth |
SparseAttentionUtils.replace_self_attention_layer_with_sparse_self_attention_layer |
4 | 0 | 0 |
meth |
SparseAttentionUtils.pad_to_block_size |
9 | 0 | 0 |
meth |
SparseAttentionUtils.unpad_sequence_output |
3 | 0 | 0 |
deepspeed.ops.sparse_attention.sparse_self_attention (28 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SparseSelfAttention.init |
5 | 0 | 0 |
meth |
SparseSelfAttention.get_layout |
2 | 0 | 0 |
meth |
SparseSelfAttention.get_ops |
3 | 0 | 0 |
meth |
SparseSelfAttention.transpose_key_for_scores |
3 | 0 | 0 |
meth |
SparseSelfAttention.transpose_mask_for_sparse |
4 | 0 | 0 |
meth |
SparseSelfAttention.forward |
7 | 0 | 0 |
attr |
SparseSelfAttention.sparsity_config |
1 | 0 | 0 |
attr |
SparseSelfAttention.key_padding_mask_mode |
1 | 0 | 0 |
attr |
SparseSelfAttention.attn_mask_mode |
1 | 0 | 0 |
attr |
SparseSelfAttention.ops |
1 | 0 | 0 |
deepspeed.ops.sparse_attention.sparsity_config (122 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DenseSparsityConfig.init |
4 | 0 | 0 |
meth |
DenseSparsityConfig.make_layout |
2 | 0 | 0 |
meth |
VariableSparsityConfig.init |
10 | 0 | 0 |
meth |
VariableSparsityConfig.set_random_layout |
3 | 0 | 0 |
meth |
VariableSparsityConfig.set_local_layout |
3 | 0 | 0 |
meth |
VariableSparsityConfig.set_global_layout |
3 | 0 | 0 |
meth |
VariableSparsityConfig.make_layout |
2 | 0 | 0 |
attr |
VariableSparsityConfig.num_random_blocks |
1 | 0 | 0 |
attr |
VariableSparsityConfig.local_window_blocks |
1 | 0 | 0 |
attr |
VariableSparsityConfig.global_block_indices |
1 | 0 | 0 |
attr |
VariableSparsityConfig.global_block_end_indices |
1 | 0 | 0 |
attr |
VariableSparsityConfig.attention |
1 | 0 | 0 |
attr |
VariableSparsityConfig.horizontal_global_attention |
1 | 0 | 0 |
meth |
BSLongformerSparsityConfig.init |
8 | 0 | 0 |
meth |
BSLongformerSparsityConfig.set_sliding_window_layout |
3 | 0 | 0 |
meth |
BSLongformerSparsityConfig.set_global_layout |
3 | 0 | 0 |
meth |
BSLongformerSparsityConfig.make_layout |
2 | 0 | 0 |
attr |
BSLongformerSparsityConfig.num_sliding_window_blocks |
1 | 0 | 0 |
attr |
BSLongformerSparsityConfig.global_block_indices |
1 | 0 | 0 |
attr |
BSLongformerSparsityConfig.attention |
1 | 0 | 0 |
attr |
BSLongformerSparsityConfig.global_block_end_indices |
1 | 0 | 0 |
meth |
SparsityConfig.init |
4 | 0 | 0 |
meth |
SparsityConfig.setup_layout |
2 | 0 | 0 |
meth |
SparsityConfig.check_and_propagate_first_head_layout |
2 | 0 | 0 |
attr |
SparsityConfig.num_heads |
1 | 0 | 0 |
attr |
SparsityConfig.block |
1 | 0 | 0 |
attr |
SparsityConfig.different_layout_per_head |
1 | 0 | 0 |
attr |
SparsityConfig.num_layout_heads |
1 | 0 | 0 |
meth |
FixedSparsityConfig.init |
9 | 0 | 0 |
meth |
FixedSparsityConfig.set_local_layout |
3 | 0 | 0 |
meth |
FixedSparsityConfig.set_global_layout |
3 | 0 | 0 |
meth |
FixedSparsityConfig.make_layout |
2 | 0 | 0 |
attr |
FixedSparsityConfig.num_local_blocks |
1 | 0 | 0 |
attr |
FixedSparsityConfig.num_global_blocks |
1 | 0 | 0 |
attr |
FixedSparsityConfig.attention |
1 | 0 | 0 |
attr |
FixedSparsityConfig.horizontal_global_attention |
1 | 0 | 0 |
attr |
FixedSparsityConfig.num_different_global_patterns |
1 | 0 | 0 |
meth |
BigBirdSparsityConfig.init |
8 | 0 | 0 |
meth |
BigBirdSparsityConfig.set_random_layout |
3 | 0 | 0 |
meth |
BigBirdSparsityConfig.set_sliding_window_layout |
3 | 0 | 0 |
meth |
BigBirdSparsityConfig.set_global_layout_itc |
3 | 0 | 0 |
meth |
BigBirdSparsityConfig.make_layout |
2 | 0 | 0 |
attr |
BigBirdSparsityConfig.num_random_blocks |
1 | 0 | 0 |
attr |
BigBirdSparsityConfig.num_sliding_window_blocks |
1 | 0 | 0 |
attr |
BigBirdSparsityConfig.num_global_blocks |
1 | 0 | 0 |
attr |
BigBirdSparsityConfig.attention |
1 | 0 | 0 |
meth |
LocalSlidingWindowSparsityConfig.init |
5 | 0 | 0 |
meth |
LocalSlidingWindowSparsityConfig.set_sliding_window_layout |
3 | 0 | 0 |
meth |
LocalSlidingWindowSparsityConfig.make_layout |
2 | 0 | 0 |
attr |
LocalSlidingWindowSparsityConfig.num_sliding_window_blocks |
1 | 0 | 0 |
attr |
LocalSlidingWindowSparsityConfig.attention |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.config (83 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TransformerConfig.init |
5 | 0 | 0 |
attr |
TransformerConfig.layer_id |
1 | 0 | 0 |
attr |
TransformerConfig.hidden_size |
1 | 0 | 0 |
attr |
TransformerConfig.intermediate_size |
1 | 0 | 0 |
attr |
TransformerConfig.heads |
1 | 0 | 0 |
attr |
TransformerConfig.num_hidden_layers |
1 | 0 | 0 |
meth |
DeepSpeedInferenceConfig.init |
37 | 0 | 0 |
meth |
DeepSpeedInferenceConfig.from_dict |
2 | 0 | 0 |
meth |
DeepSpeedInferenceConfig.from_json_file |
2 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.dtype |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.pre_layer_norm |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.norm_type |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.local_rank |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.stochastic_mode |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.epsilon |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.mp_size |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.scale_attention |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.triangular_masking |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.local_attention |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.window_size |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.rotary_dim |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.rotate_half |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.rotate_every_two |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.return_tuple |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.mlp_after_attn |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.mlp_act_func_type |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.training_mp_size |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.bigscience_bloom |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.max_out_tokens |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.min_out_tokens |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.scale_attn_by_inverse_layer_idx |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.enable_qkv_quantization |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.use_mup |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.return_single_tuple |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.set_empty_params |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.transposed_mode |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.use_triton |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.triton_autotune |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.num_kv |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.rope_theta |
1 | 0 | 0 |
attr |
DeepSpeedInferenceConfig.invert_mask |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.diffusers_2d_transformer (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Diffusers2DTransformerConfig.init |
2 | 0 | 0 |
attr |
Diffusers2DTransformerConfig.int8_quantization |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.diffusers_attention (55 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedDiffusersAttention.init |
2 | 0 | 0 |
meth |
DeepSpeedDiffusersAttention.allocate_workspace |
2 | 0 | 0 |
meth |
DeepSpeedDiffusersAttention.forward |
4 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.config |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_qkvw |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_kw |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_vw |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_qw |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_qkvb |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_ow |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.attn_ob |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.do_out_bias |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.triton_flash_attn_kernel |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.num_attention_heads_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.hidden_size_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.hidden_size_per_attention_head |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.norm_factor |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.workspace |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.score_context_func |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.linear_func |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersAttention.pad_transform_func |
1 | 0 | 0 |
func |
load_triton_flash_attn |
1 | 0 | 0 |
meth |
DeepSpeedDiffusersAttentionFunction.forward |
22 | 0 | 0 |
meth |
DeepSpeedDiffusersAttentionFunction.backward |
6 | 0 | 0 |
deepspeed.ops.transformer.inference.diffusers_transformer_block (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedDiffusersTransformerBlock.init |
3 | 2 | 0 |
meth |
DeepSpeedDiffusersTransformerBlock.forward |
5 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.quantizer |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.config |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.ff1_w |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.ff1_b |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.ff2_w |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.ff2_b |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm1_g |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm1_b |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm1_eps |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm2_g |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm2_b |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm2_eps |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm3_g |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm3_b |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.norm3_eps |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.attn_1 |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.attn_2 |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.gated_activation |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.layer_norm |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.attn_1_bias |
1 | 0 | 0 |
attr |
DeepSpeedDiffusersTransformerBlock.attn_2_bias |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.ds_attention (70 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedSelfAttention.init |
6 | 0 | 0 |
meth |
DeepSpeedSelfAttention.compute_attention |
8 | 0 | 0 |
meth |
DeepSpeedSelfAttention._merge_qkv |
1 | 0 | 0 |
meth |
DeepSpeedSelfAttention.forward |
13 | 0 | 0 |
attr |
DeepSpeedSelfAttention.config |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.num_attention_heads_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.num_kv_partition |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.hidden_size_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.hidden_size_per_attention_head |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.mp_group |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.q_scales |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.q_groups |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.merge_count |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.norm_factor |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.qkv_func |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.score_context_func |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.linear_func |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.vector_matmul_func |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_qw |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_qb |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_kw |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_kb |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_vw |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_vb |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_qkvw |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_qkvb |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_ow |
1 | 0 | 0 |
attr |
DeepSpeedSelfAttention.attn_ob |
1 | 0 | 0 |
meth |
BloomSelfAttention.init |
3 | 0 | 0 |
meth |
BloomSelfAttention._transpose_for_context |
2 | 0 | 0 |
meth |
BloomSelfAttention._split_tensor_along_last_dim |
4 | 0 | 0 |
meth |
BloomSelfAttention.compute_attention |
8 | 0 | 0 |
attr |
BloomSelfAttention.softmax_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.ds_mlp (34 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedMLP.init |
7 | 0 | 0 |
meth |
DeepSpeedMLP._merge_inter_w |
1 | 0 | 0 |
meth |
DeepSpeedMLP.forward |
5 | 0 | 0 |
attr |
DeepSpeedMLP.config |
1 | 0 | 0 |
attr |
DeepSpeedMLP.intm_w_sz_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedMLP.intm_o_sz_per_partition |
1 | 0 | 0 |
attr |
DeepSpeedMLP.q_scales |
1 | 0 | 0 |
attr |
DeepSpeedMLP.q_groups |
1 | 0 | 0 |
attr |
DeepSpeedMLP.merge_count |
1 | 0 | 0 |
attr |
DeepSpeedMLP.mp_group |
1 | 0 | 0 |
attr |
DeepSpeedMLP.mlp_gemm_func |
1 | 0 | 0 |
attr |
DeepSpeedMLP.vector_matmul_func |
1 | 0 | 0 |
attr |
DeepSpeedMLP.fused_gemm_gelu |
1 | 0 | 0 |
attr |
DeepSpeedMLP.residual_add_func |
1 | 0 | 0 |
attr |
DeepSpeedMLP.attn_nw |
1 | 0 | 0 |
attr |
DeepSpeedMLP.attn_nb |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_w |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_b |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_up_w |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_up_b |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_gate_w |
1 | 0 | 0 |
attr |
DeepSpeedMLP.inter_gate_b |
1 | 0 | 0 |
attr |
DeepSpeedMLP.output_w |
1 | 0 | 0 |
attr |
DeepSpeedMLP.output_b |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.moe_inference (140 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedMoEInferenceConfig.init |
29 | 0 | 0 |
meth |
DeepSpeedMoEInferenceConfig.from_dict |
2 | 0 | 0 |
meth |
DeepSpeedMoEInferenceConfig.from_json_file |
2 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.moe_experts |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.k |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.capacity_factor |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.eval_capacity_factor |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.min_capacity |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.noisy_gate_policy |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.drop_tokens |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.use_rts |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.global_experts |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.mlp_type |
1 | 0 | 0 |
attr |
DeepSpeedMoEInferenceConfig.scale_attn_by_inverse_layer_idx |
1 | 0 | 0 |
meth |
DeepSpeedMLPFunction.forward |
15 | 0 | 0 |
meth |
DeepSpeedMLPFunction.backward |
3 | 0 | 0 |
meth |
DeepSpeedMoEMLP.init |
7 | 0 | 0 |
meth |
DeepSpeedMoEMLP.forward |
3 | 0 | 0 |
attr |
DeepSpeedMoEMLP.config |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.attn_nw |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.attn_nb |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.inter_w |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.inter_b |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.output_w |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.output_b |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.q_scales |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.q_groups |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.merge_count |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.mp_group |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.gelu_gemm_func |
1 | 0 | 0 |
attr |
DeepSpeedMoEMLP.vector_matmul_func |
1 | 0 | 0 |
meth |
DeepSpeedMoEInference.init |
9 | 0 | 0 |
meth |
DeepSpeedMoEInference.res_coef_func |
3 | 0 | 0 |
meth |
DeepSpeedMoEInference.moe_gate_einsum |
2 | 0 | 0 |
meth |
DeepSpeedMoEInference.expert_exec |
2 | 0 | 0 |
meth |
DeepSpeedMoEInference._alltoall |
2 | 0 | 0 |
meth |
DeepSpeedMoEInference.scale_expert_output |
4 | 0 | 0 |
meth |
DeepSpeedMoEInference.forward |
14 | 0 | 0 |
attr |
DeepSpeedMoEInference.config |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.attention |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.attn_nw |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.attn_nb |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.norm_w |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.norm_b |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.mlp |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.moe_gate |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.ep_group |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.mp_group |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.expert_mp_group |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.bias_residual_func |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.ds_layernorm |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.einsum_sec_sm_ecm |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.moe_res_matmul |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.res_mlp |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.res_coef |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.coef_func |
1 | 0 | 0 |
attr |
DeepSpeedMoEInference.vector_matmul_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.base (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BaseOp.init |
2 | 1 | 0 |
attr |
BaseOp.config |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.bias_add (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BiasAddOp.init |
2 | 1 | 0 |
meth |
BiasAddOp.bias_add_fallback |
3 | 0 | 0 |
meth |
BiasAddOp.forward |
3 | 2 | 0 |
attr |
BiasAddOp.bias_add_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.bias_gelu (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BiasGeluOp.init |
2 | 1 | 0 |
meth |
BiasGeluOp.bias_gelu_fallback |
3 | 0 | 0 |
meth |
BiasGeluOp.forward |
3 | 2 | 0 |
attr |
BiasGeluOp.bias_gelu_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.bias_relu (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BiasReluOp.init |
2 | 1 | 0 |
meth |
BiasReluOp.bias_relu_fallback |
3 | 0 | 0 |
meth |
BiasReluOp.forward |
3 | 2 | 0 |
attr |
BiasReluOp.bias_relu_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.bias_residual (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BiasResidualOp.init |
2 | 1 | 0 |
meth |
BiasResidualOp.bias_residual_fallback |
4 | 0 | 0 |
meth |
BiasResidualOp.forward |
4 | 0 | 0 |
attr |
BiasResidualOp.bias_residual_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.einsum_sec_sm_ecm (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
EinsumSecSmEcmOp.init |
2 | 1 | 0 |
meth |
EinsumSecSmEcmOp.einsum_sec_sm_ecm_fallback |
3 | 0 | 0 |
meth |
EinsumSecSmEcmOp.forward |
3 | 0 | 0 |
attr |
EinsumSecSmEcmOp.einsum_sec_sm_ecm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.gated_activation (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GatedActivationOp.init |
2 | 1 | 0 |
meth |
GatedActivationOp.gated_activation_fallback |
4 | 0 | 0 |
meth |
GatedActivationOp.forward |
4 | 3 | 0 |
attr |
GatedActivationOp.gated_activation_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.gelu_gemm (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
GELUGemmOp.init |
2 | 1 | 0 |
meth |
GELUGemmOp.gelu_gemm_fallback |
9 | 0 | 0 |
meth |
GELUGemmOp.forward |
5 | 4 | 0 |
attr |
GELUGemmOp.fused_gemm_gelu |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.layer_norm (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
LayerNormOp.init |
2 | 1 | 0 |
meth |
LayerNormOp.layer_norm_residual |
7 | 0 | 0 |
meth |
LayerNormOp.layer_norm_residual_store_pre_ln_res |
7 | 0 | 0 |
meth |
LayerNormOp.layer_norm_fallback |
5 | 0 | 0 |
meth |
LayerNormOp.forward |
5 | 0 | 0 |
attr |
LayerNormOp.layer_norm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.linear (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
LinearOp.init |
2 | 1 | 0 |
meth |
LinearOp.linear_fallback |
9 | 0 | 0 |
meth |
LinearOp.forward |
9 | 8 | 0 |
meth |
LinearOp._triton_autotune |
5 | 0 | 0 |
attr |
LinearOp.linear_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.mlp_gemm (33 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MLPGemmOp.init |
2 | 1 | 0 |
meth |
MLPGemmOp.mlp_gemm_fallback |
17 | 0 | 0 |
meth |
MLPGemmOp.rms_mlp_gemm_fallback |
12 | 0 | 0 |
meth |
MLPGemmOp.forward |
9 | 8 | 0 |
attr |
MLPGemmOp.pre_rms_norm |
1 | 0 | 0 |
attr |
MLPGemmOp.mlp_gemm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.moe_res_matmul (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MoEResMatmulOp.init |
2 | 1 | 0 |
meth |
MoEResMatmulOp.moe_res_matmul_fallback |
4 | 0 | 0 |
meth |
MoEResMatmulOp.forward |
4 | 0 | 0 |
attr |
MoEResMatmulOp.moe_res_matmul_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.pad_transform (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PadTransformOp.init |
2 | 1 | 0 |
meth |
PadTransformOp.pad_transform_fallback |
6 | 0 | 0 |
meth |
PadTransformOp.forward |
6 | 0 | 0 |
attr |
PadTransformOp.pad_transform_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.pre_rms_norm (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PreRMSNormOp.init |
2 | 1 | 0 |
meth |
PreRMSNormOp.pre_rms_norm_fallback |
5 | 0 | 0 |
meth |
PreRMSNormOp.forward |
5 | 0 | 0 |
attr |
PreRMSNormOp.pre_rms_norm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.qkv_gemm (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
QKVGemmOp.init |
2 | 1 | 0 |
meth |
QKVGemmOp._triton_autotune |
5 | 0 | 0 |
meth |
QKVGemmOp.qkv_gemm_fallback |
11 | 0 | 0 |
meth |
QKVGemmOp.rms_qkv_gemm_fallback |
8 | 0 | 0 |
meth |
QKVGemmOp.forward |
6 | 5 | 0 |
attr |
QKVGemmOp.qkv_gemm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.residual_add (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ResidualAddOp.init |
2 | 1 | 0 |
meth |
ResidualAddOp.res_add_bias |
8 | 0 | 0 |
meth |
ResidualAddOp.residual_add_fallback |
10 | 0 | 0 |
meth |
ResidualAddOp.forward |
8 | 7 | 0 |
attr |
ResidualAddOp.vector_add |
1 | 0 | 0 |
attr |
ResidualAddOp.residual_add_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.rms_norm (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RMSNormOp.init |
2 | 1 | 0 |
meth |
RMSNormOp.rms_norm_fallback |
4 | 0 | 0 |
meth |
RMSNormOp.forward |
4 | 0 | 0 |
attr |
RMSNormOp.rms_norm_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.softmax (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SoftmaxOp.init |
2 | 1 | 0 |
meth |
SoftmaxOp.softmax_fallback |
12 | 0 | 0 |
meth |
SoftmaxOp.forward |
11 | 10 | 0 |
attr |
SoftmaxOp.num_attention_heads_per_partition |
1 | 0 | 0 |
attr |
SoftmaxOp.softmax_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.softmax_context (32 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SoftmaxContextOp.init |
2 | 1 | 0 |
meth |
SoftmaxContextOp.transform4d_0213 |
3 | 0 | 0 |
meth |
SoftmaxContextOp.bias_add_transform_0213 |
6 | 0 | 0 |
meth |
SoftmaxContextOp.softmax_context_fallback |
20 | 0 | 0 |
meth |
SoftmaxContextOp.forward |
13 | 12 | 0 |
attr |
SoftmaxContextOp.softmax_context_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.vector_add (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
VectorAddOp.init |
2 | 1 | 0 |
meth |
VectorAddOp.vector_add_fallback |
4 | 0 | 0 |
meth |
VectorAddOp.forward |
4 | 0 | 0 |
attr |
VectorAddOp.vector_add_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.vector_matmul (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
VectorMatMulOp.init |
2 | 1 | 0 |
meth |
VectorMatMulOp.vector_matmul_fallback |
7 | 0 | 0 |
meth |
VectorMatMulOp.forward |
4 | 3 | 0 |
meth |
VectorMatMulOp._triton_autotune |
5 | 0 | 0 |
attr |
VectorMatMulOp.vector_matmul_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.op_binding.workspace (91 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
InferenceContext.init |
1 | 0 | 0 |
meth |
InferenceContext.Instance |
1 | 0 | 0 |
meth |
InferenceContext.gen_workspace |
12 | 0 | 0 |
meth |
InferenceContext.retake_workspace |
1 | 0 | 0 |
meth |
InferenceContext._retake_workspace |
1 | 0 | 0 |
meth |
InferenceContext.update_cache |
6 | 0 | 0 |
meth |
InferenceContext.release_workspace |
1 | 0 | 0 |
meth |
InferenceContext.reset_tokens |
2 | 0 | 0 |
meth |
InferenceContext.current_tokens |
1 | 0 | 0 |
meth |
InferenceContext.advance_tokens |
1 | 0 | 0 |
meth |
InferenceContext.get_kv_cache |
1 | 0 | 0 |
meth |
InferenceContext.get_rotary |
4 | 0 | 0 |
meth |
InferenceContext.get_max_tokens_num |
1 | 0 | 0 |
attr |
InferenceContext.kv_cache |
1 | 0 | 0 |
attr |
InferenceContext.kv_cache_elem_dtype |
1 | 0 | 0 |
attr |
InferenceContext.num_tokens |
1 | 0 | 0 |
attr |
InferenceContext.kv_cache_num_layers |
1 | 0 | 0 |
attr |
InferenceContext.kv_cache_size |
1 | 0 | 0 |
attr |
InferenceContext.max_out_tokens |
1 | 0 | 0 |
attr |
InferenceContext.rotary |
1 | 0 | 0 |
attr |
InferenceContext.allocate_called |
1 | 0 | 0 |
attr |
InferenceContext.static_shapes |
1 | 0 | 0 |
meth |
WorkspaceOp.init |
2 | 1 | 0 |
meth |
WorkspaceOp.allocate_workspace |
3 | 0 | 0 |
meth |
WorkspaceOp.release_workspace |
1 | 0 | 0 |
meth |
WorkspaceOp.reset_cache |
1 | 0 | 0 |
meth |
WorkspaceOp.retake_workspace |
1 | 0 | 0 |
meth |
WorkspaceOp.allocate_workspace_fp32_fallback |
11 | 0 | 0 |
meth |
WorkspaceOp.allocate_workspace_bf16_fallback |
11 | 0 | 0 |
meth |
WorkspaceOp.allocate_workspace_fp16_fallback |
11 | 0 | 0 |
meth |
WorkspaceOp.reset_cache_fallback |
1 | 0 | 0 |
meth |
WorkspaceOp.release_workspace_fallback |
1 | 0 | 0 |
meth |
WorkspaceOp.retake_workspace_fallback |
1 | 0 | 0 |
meth |
WorkspaceOp.is_allocated |
1 | 0 | 0 |
attr |
WorkspaceOp.inference_context |
1 | 0 | 0 |
attr |
WorkspaceOp.release_workspace_func |
1 | 0 | 0 |
attr |
WorkspaceOp.retake_workspace_func |
1 | 0 | 0 |
attr |
WorkspaceOp.reset_cache_func |
1 | 0 | 0 |
attr |
WorkspaceOp.allocate_workspace_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.triton.attention (65 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TritonSelfAttention.init |
7 | 0 | 0 |
meth |
TritonSelfAttention._triton_autotune |
8 | 0 | 0 |
meth |
TritonSelfAttention.ds_compute_attention |
8 | 0 | 0 |
meth |
TritonSelfAttention.forward |
14 | 0 | 0 |
attr |
TritonSelfAttention.config |
1 | 0 | 0 |
attr |
TritonSelfAttention.num_attention_heads_per_partition |
1 | 0 | 0 |
attr |
TritonSelfAttention.hidden_size_per_partition |
1 | 0 | 0 |
attr |
TritonSelfAttention.hidden_size_per_attention_head |
1 | 0 | 0 |
attr |
TritonSelfAttention.mp_group |
1 | 0 | 0 |
attr |
TritonSelfAttention.use_flash |
1 | 0 | 0 |
attr |
TritonSelfAttention.q_scales |
1 | 0 | 0 |
attr |
TritonSelfAttention.q_groups |
1 | 0 | 0 |
attr |
TritonSelfAttention.merge_count |
1 | 0 | 0 |
attr |
TritonSelfAttention.norm_factor |
1 | 0 | 0 |
attr |
TritonSelfAttention.qkv_func |
1 | 0 | 0 |
attr |
TritonSelfAttention.score_context_func |
1 | 0 | 0 |
attr |
TritonSelfAttention.linear_func |
1 | 0 | 0 |
attr |
TritonSelfAttention.vector_matmul_func |
1 | 0 | 0 |
attr |
TritonSelfAttention.hidden_size |
1 | 0 | 0 |
attr |
TritonSelfAttention.head_size |
1 | 0 | 0 |
attr |
TritonSelfAttention.scale |
1 | 0 | 0 |
attr |
TritonSelfAttention.triangular_masking |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_qw |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_qb |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_kw |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_kb |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_vw |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_vb |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_qkvw |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_qkvb |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_ow |
1 | 0 | 0 |
attr |
TritonSelfAttention.attn_ob |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.triton.gelu (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
gelu_functor |
2 | 0 | 0 |
func |
gelu_kernel |
5 | 1 | 0 |
deepspeed.ops.transformer.inference.triton.layer_norm (41 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
layer_norm_kernel |
9 | 1 | 0 |
func |
layer_norm_residual |
7 | 0 | 0 |
func |
layer_norm |
5 | 0 | 0 |
func |
layer_norm_residual_kernel |
11 | 1 | 0 |
func |
layer_norm_residual_bias_kernel |
12 | 1 | 0 |
deepspeed.ops.transformer.inference.triton.matmul_ext (90 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
is_nfs_path |
2 | 0 | 0 |
meth |
Fp16Matmul.init |
2 | 0 | 0 |
meth |
Fp16Matmul.skip_autotune |
1 | 0 | 0 |
meth |
Fp16Matmul.forward |
6 | 0 | 0 |
meth |
Fp16Matmul._matmul_4d |
3 | 0 | 0 |
meth |
Fp16Matmul._score_4d_matmul |
5 | 0 | 0 |
meth |
Fp16Matmul._context_4d_matmul |
4 | 0 | 0 |
meth |
Fp16Matmul._ref_forward |
6 | 0 | 0 |
meth |
Fp16Matmul._check_parity |
12 | 0 | 0 |
meth |
Fp16Matmul._read_autotune_table |
1 | 0 | 0 |
meth |
Fp16Matmul._write_autotune_table |
1 | 0 | 0 |
meth |
Fp16Matmul._update_autotune_table |
1 | 0 | 0 |
meth |
AutotuneCacheManager.init |
2 | 0 | 0 |
meth |
AutotuneCacheManager.has_file |
1 | 0 | 0 |
meth |
AutotuneCacheManager.put |
2 | 0 | 0 |
meth |
AutotuneCacheManager.load |
1 | 0 | 0 |
attr |
AutotuneCacheManager.key |
1 | 0 | 0 |
attr |
AutotuneCacheManager.file_path |
1 | 0 | 0 |
attr |
AutotuneCacheManager.lock_path |
1 | 0 | 0 |
attr |
AutotuneCacheManager.cache_dir |
1 | 0 | 0 |
attr |
fp16_matmul |
1 | 0 | 0 |
func |
matmul_ext_update_autotune_table |
1 | 0 | 0 |
meth |
MatmulExt.forward |
7 | 0 | 0 |
meth |
TritonCacheDir.warn_if_nfs |
2 | 0 | 0 |
meth |
TritonCacheDir.default_cache_dir |
1 | 0 | 0 |
meth |
TritonMatmul.init |
1 | 0 | 0 |
meth |
TritonMatmul._ref_forward |
4 | 0 | 0 |
meth |
TritonMatmul._read_autotune_table |
3 | 0 | 0 |
meth |
TritonMatmul._write_autotune_table |
3 | 0 | 0 |
meth |
TritonMatmul._update_autotune_table |
3 | 0 | 0 |
meth |
TritonMatmul.forward |
6 | 0 | 0 |
func |
bias_add_activation |
4 | 0 | 0 |
deepspeed.ops.transformer.inference.triton.mlp (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TritonMLP.init |
7 | 0 | 0 |
meth |
TritonMLP.forward |
5 | 0 | 0 |
attr |
TritonMLP.config |
1 | 0 | 0 |
attr |
TritonMLP.attn_nw |
1 | 0 | 0 |
attr |
TritonMLP.attn_nb |
1 | 0 | 0 |
attr |
TritonMLP.inter_w |
1 | 0 | 0 |
attr |
TritonMLP.inter_b |
1 | 0 | 0 |
attr |
TritonMLP.output_w |
1 | 0 | 0 |
attr |
TritonMLP.output_b |
1 | 0 | 0 |
attr |
TritonMLP.q_scales |
1 | 0 | 0 |
attr |
TritonMLP.q_groups |
1 | 0 | 0 |
attr |
TritonMLP.merge_count |
1 | 0 | 0 |
attr |
TritonMLP.mp_group |
1 | 0 | 0 |
attr |
TritonMLP.mlp_gemm_func |
1 | 0 | 0 |
attr |
TritonMLP.vector_matmul_func |
1 | 0 | 0 |
attr |
TritonMLP.fused_gemm_gelu |
1 | 0 | 0 |
attr |
TritonMLP.residual_add_func |
1 | 0 | 0 |
deepspeed.ops.transformer.inference.triton.ops (58 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
linear_func |
8 | 0 | 0 |
func |
qkv_gemm_func |
12 | 0 | 0 |
func |
vector_matmul_func |
7 | 0 | 0 |
func |
fused_gemm_gelu |
13 | 0 | 0 |
func |
mlp_gemm_func |
18 | 0 | 0 |
deepspeed.ops.transformer.inference.triton.residual_add (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
residual_add_bias |
10 | 9 | 0 |
func |
residual_add_bias_kernel |
14 | 5 | 0 |
deepspeed.ops.transformer.inference.triton.softmax (13 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
softmax |
4 | 3 | 0 |
func |
masked_softmax_kernel |
8 | 1 | 0 |
func |
softmax_kernel |
6 | 1 | 0 |
deepspeed.ops.transformer.inference.triton.triton_matmul_kernel (27 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
matmul_4d_kernel |
28 | 5 | 0 |
func |
matmul_4d_prune_config |
4 | 0 | 0 |
deepspeed.ops.transformer.inference.triton_ops (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
triton_flash_attn.init |
1 | 0 | 0 |
meth |
triton_flash_attn.forward |
6 | 0 | 0 |
deepspeed.ops.transformer.transformer (109 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedTransformerConfig.init |
21 | 0 | 0 |
meth |
DeepSpeedTransformerConfig.from_dict |
2 | 0 | 0 |
meth |
DeepSpeedTransformerConfig.from_json_file |
2 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.fp16 |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.pre_layer_norm |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.local_rank |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.seed |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.normalize_invertible |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.gelu_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.adjust_init_range |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.test_gemm |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.layer_norm_eps |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.training |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.is_grad_enabled |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.attn_dropout_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.stochastic_mode |
1 | 0 | 0 |
attr |
DeepSpeedTransformerConfig.return_tuple |
1 | 0 | 0 |
meth |
DeepSpeedTransformerFunction.forward |
20 | 0 | 0 |
meth |
DeepSpeedTransformerFunction.backward |
3 | 0 | 0 |
meth |
TransformerConfig.init |
9 | 0 | 0 |
attr |
TransformerConfig.layer_id |
1 | 0 | 0 |
attr |
TransformerConfig.batch_size |
1 | 0 | 0 |
attr |
TransformerConfig.hidden_size |
1 | 0 | 0 |
attr |
TransformerConfig.intermediate_size |
1 | 0 | 0 |
attr |
TransformerConfig.heads |
1 | 0 | 0 |
attr |
TransformerConfig.attn_dropout_ratio |
1 | 0 | 0 |
attr |
TransformerConfig.hidden_dropout_ratio |
1 | 0 | 0 |
attr |
TransformerConfig.num_hidden_layers |
1 | 0 | 0 |
attr |
TransformerConfig.initializer_range |
1 | 0 | 0 |
meth |
DeepSpeedTransformerLayer.init |
4 | 0 | 0 |
meth |
DeepSpeedTransformerLayer.init_transformer_weights |
2 | 0 | 0 |
meth |
DeepSpeedTransformerLayer.forward |
10 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.config |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_qkvw |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_qkvb |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_ow |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_ob |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_nw |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.attn_nb |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.inter_w |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.inter_b |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.output_w |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.output_b |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.norm_w |
1 | 0 | 0 |
attr |
DeepSpeedTransformerLayer.norm_b |
1 | 0 | 0 |
deepspeed.profiling.config (9 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedFlopsProfilerConfig.init |
2 | 0 | 0 |
meth |
DeepSpeedFlopsProfilerConfig._initialize |
2 | 0 | 0 |
attr |
DeepSpeedFlopsProfilerConfig.enabled |
1 | 0 | 0 |
attr |
DeepSpeedFlopsProfilerConfig.recompute_fwd_factor |
1 | 0 | 0 |
attr |
DeepSpeedFlopsProfilerConfig.profile_step |
1 | 0 | 0 |
attr |
DeepSpeedFlopsProfilerConfig.module_depth |
1 | 0 | 0 |
attr |
DeepSpeedFlopsProfilerConfig.top_modules |
1 | 0 | 0 |
deepspeed.profiling.flops_profiler.profiler (79 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FlopsProfiler.init |
4 | 0 | 0 |
meth |
FlopsProfiler.start_profile |
2 | 0 | 0 |
meth |
FlopsProfiler.stop_profile |
1 | 0 | 0 |
meth |
FlopsProfiler.reset_profile |
1 | 0 | 0 |
meth |
FlopsProfiler.end_profile |
1 | 0 | 0 |
meth |
FlopsProfiler.get_total_flops |
2 | 0 | 0 |
meth |
FlopsProfiler.get_total_macs |
2 | 0 | 0 |
meth |
FlopsProfiler.get_total_duration |
2 | 0 | 0 |
meth |
FlopsProfiler.get_total_params |
2 | 0 | 0 |
meth |
FlopsProfiler.is_expert_tensor_parallelism_enabled |
1 | 0 | 0 |
meth |
FlopsProfiler.print_model_profile |
6 | 0 | 0 |
meth |
FlopsProfiler.print_model_aggregated_profile |
3 | 0 | 0 |
attr |
FlopsProfiler.model |
1 | 0 | 0 |
attr |
FlopsProfiler.ds_engine |
1 | 0 | 0 |
attr |
FlopsProfiler.recompute_fwd_factor |
1 | 0 | 0 |
attr |
FlopsProfiler.started |
1 | 0 | 0 |
attr |
FlopsProfiler.func_patched |
1 | 0 | 0 |
func |
get_module_duration |
2 | 0 | 0 |
func |
get_model_profile |
14 | 0 | 0 |
func |
bytes_to_string |
4 | 0 | 0 |
func |
duration_to_string |
4 | 0 | 0 |
func |
get_module_flops |
2 | 0 | 0 |
func |
number_to_string |
4 | 0 | 0 |
func |
get_module_macs |
2 | 0 | 0 |
func |
macs_to_string |
4 | 0 | 0 |
func |
flops_to_string |
4 | 0 | 0 |
func |
wrapFunc |
3 | 0 | 0 |
func |
params_to_string |
4 | 0 | 0 |
deepspeed.runtime.activation_checkpointing.checkpointing (75 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_cuda_rng_tracker |
1 | 0 | 0 |
func |
get_partition_size |
2 | 0 | 0 |
func |
merge_tensors |
4 | 0 | 0 |
func |
gather_partitioned_activations |
3 | 0 | 0 |
func |
checkpoint |
3 | 0 | 0 |
func |
partition_activations_in_checkpoint |
2 | 0 | 0 |
func |
get_partition_start |
2 | 0 | 0 |
func |
model_parallel_reconfigure_tp_seed |
2 | 0 | 0 |
func |
set_num_layers |
2 | 0 | 0 |
func |
detach_variable |
3 | 0 | 0 |
func |
non_reentrant_checkpoint |
3 | 0 | 0 |
func |
get_partitioned_activations_for_backward |
4 | 0 | 0 |
func |
configure |
9 | 0 | 0 |
func |
partition_activations |
4 | 0 | 0 |
func |
get_cpu_activations_for_backward |
3 | 0 | 0 |
meth |
CheckpointFunction.forward |
5 | 0 | 0 |
meth |
CheckpointFunction.backward |
3 | 0 | 0 |
func |
is_configured |
1 | 0 | 0 |
func |
extract_tensors |
2 | 0 | 0 |
func |
model_parallel_cuda_manual_seed |
2 | 0 | 0 |
meth |
CudaRNGStatesTracker.init |
1 | 0 | 0 |
meth |
CudaRNGStatesTracker.reset |
1 | 0 | 0 |
meth |
CudaRNGStatesTracker.get_states |
1 | 0 | 0 |
meth |
CudaRNGStatesTracker.set_states |
2 | 0 | 0 |
meth |
CudaRNGStatesTracker.add |
3 | 0 | 0 |
meth |
CudaRNGStatesTracker.fork |
2 | 0 | 0 |
attr |
CudaRNGStatesTracker.states_ |
1 | 0 | 0 |
attr |
CudaRNGStatesTracker.seeds_ |
1 | 0 | 0 |
func |
is_activation_to_checkpoint |
2 | 0 | 0 |
func |
reset |
1 | 0 | 0 |
deepspeed.runtime.activation_checkpointing.config (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedActivationCheckpointingConfig.init |
2 | 0 | 0 |
meth |
DeepSpeedActivationCheckpointingConfig._initialize |
2 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.partition_activations |
1 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.contiguous_memory_optimization |
1 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.cpu_checkpointing |
1 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.number_checkpoints |
1 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.synchronize_checkpoint_boundary |
1 | 0 | 0 |
attr |
DeepSpeedActivationCheckpointingConfig.profile |
1 | 0 | 0 |
deepspeed.runtime.base_optimizer (66 missing, 2 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZeROOptimizer.init |
1 | 0 | 0 |
meth |
ZeROOptimizer.report_ipg_memory_usage |
4 | 0 | 0 |
meth |
ZeROOptimizer.get_param_comm_dtype |
2 | 0 | 0 |
meth |
ZeROOptimizer.scale_if_loss |
2 | 2 | 2 |
meth |
ZeROOptimizer.backward_prologue |
1 | 0 | 0 |
meth |
ZeROOptimizer.backward_epilogue |
2 | 0 | 0 |
meth |
ZeROOptimizer.backward |
3 | 0 | 0 |
meth |
ZeROOptimizer.register_grad_acc_post_hook |
2 | 0 | 0 |
meth |
ZeROOptimizer.unregister_grad_acc_post_hooks |
1 | 0 | 0 |
meth |
ZeROOptimizer.run_grad_acc_post_hooks |
1 | 0 | 0 |
meth |
ZeROOptimizer.enter_backward |
1 | 0 | 0 |
meth |
ZeROOptimizer.exit_backward |
1 | 0 | 0 |
meth |
ZeROOptimizer.clear_backward_seen_flag |
1 | 0 | 0 |
meth |
ZeROOptimizer.should_refresh_expected_hook_count |
1 | 0 | 0 |
meth |
ZeROOptimizer.reenter_backward_if_needed |
1 | 0 | 0 |
meth |
ZeROOptimizer.update_hook_state_and_maybe_run_epilogue |
2 | 0 | 0 |
meth |
ZeROOptimizer.queue_post_backward_callback |
1 | 0 | 0 |
meth |
ZeROOptimizer._configure_master_weights |
6 | 0 | 0 |
prop |
ZeROOptimizer._remaining_grad_acc_hooks |
2 | 0 | 0 |
prop |
ZeROOptimizer._backward_active_depth |
2 | 0 | 0 |
prop |
ZeROOptimizer._backward_seen_this_step |
2 | 0 | 0 |
prop |
ZeROOptimizer._epilogue_ran_this_backward |
2 | 0 | 0 |
prop |
ZeROOptimizer._hooks_fired_this_backward |
2 | 0 | 0 |
prop |
ZeROOptimizer._max_expected_hooks_seen |
2 | 0 | 0 |
prop |
ZeROOptimizer._grad_acc_post_hooks |
2 | 0 | 0 |
meth |
BackwardHookStateManager.init |
1 | 0 | 0 |
meth |
BackwardHookStateManager.register_grad_acc_post_hook |
2 | 0 | 0 |
meth |
BackwardHookStateManager.unregister_grad_acc_post_hooks |
1 | 0 | 0 |
meth |
BackwardHookStateManager.run_grad_acc_post_hooks |
1 | 0 | 0 |
meth |
BackwardHookStateManager.enter_backward |
1 | 0 | 0 |
meth |
BackwardHookStateManager.exit_backward |
1 | 0 | 0 |
meth |
BackwardHookStateManager.reset_for_new_step |
1 | 0 | 0 |
meth |
BackwardHookStateManager.should_refresh_expected_hook_count |
1 | 0 | 0 |
meth |
BackwardHookStateManager.reenter_backward_if_needed |
1 | 0 | 0 |
meth |
BackwardHookStateManager.queue_post_backward_callback |
1 | 0 | 0 |
meth |
BackwardHookStateManager.update_hook_state_and_maybe_run_epilogue |
2 | 0 | 0 |
attr |
BackwardHookStateManager.remaining_grad_acc_hooks |
1 | 0 | 0 |
attr |
BackwardHookStateManager.backward_active_depth |
1 | 0 | 0 |
attr |
BackwardHookStateManager.backward_seen_this_step |
1 | 0 | 0 |
attr |
BackwardHookStateManager.epilogue_ran_this_backward |
1 | 0 | 0 |
attr |
BackwardHookStateManager.hooks_fired_this_backward |
1 | 0 | 0 |
attr |
BackwardHookStateManager.max_expected_hooks_seen |
1 | 0 | 0 |
attr |
BackwardHookStateManager.post_backward_callback_queued |
1 | 0 | 0 |
attr |
BackwardHookStateManager.post_backward_callback_graph_task_id |
1 | 0 | 0 |
deepspeed.runtime.bf16_optimizer (118 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
BF16_Optimizer.init |
13 | 0 | 0 |
meth |
BF16_Optimizer.destroy |
1 | 0 | 0 |
meth |
BF16_Optimizer._configure_moe_settings |
1 | 0 | 0 |
meth |
BF16_Optimizer._setup_for_real_optimizer |
1 | 0 | 0 |
meth |
BF16_Optimizer._enable_universal_checkpoint |
1 | 0 | 0 |
meth |
BF16_Optimizer._create_param_mapping |
1 | 0 | 0 |
meth |
BF16_Optimizer._link_all_hp_params |
1 | 0 | 0 |
meth |
BF16_Optimizer._lazy_init_hp_params_optimizer_state |
1 | 0 | 0 |
meth |
BF16_Optimizer._split_flat_tensor |
3 | 0 | 0 |
meth |
BF16_Optimizer._update_storage_to_flattened_tensor |
3 | 0 | 0 |
meth |
BF16_Optimizer._flatten_dense_tensors_aligned |
3 | 0 | 0 |
meth |
BF16_Optimizer.step |
2 | 0 | 0 |
meth |
BF16_Optimizer.backward_prologue |
1 | 0 | 0 |
meth |
BF16_Optimizer.backward_epilogue |
4 | 0 | 0 |
meth |
BF16_Optimizer._update_hp_grad |
5 | 0 | 0 |
meth |
BF16_Optimizer._update_hp_grads_func |
2 | 0 | 0 |
meth |
BF16_Optimizer.update_hp_grads |
2 | 0 | 0 |
meth |
BF16_Optimizer.get_grads_for_reduction |
1 | 0 | 0 |
meth |
BF16_Optimizer.get_grads_for_norm |
2 | 0 | 0 |
meth |
BF16_Optimizer.update_lp_params |
1 | 0 | 0 |
meth |
BF16_Optimizer.clear_hp_grads |
1 | 0 | 0 |
meth |
BF16_Optimizer.clear_lp_grads |
2 | 0 | 0 |
meth |
BF16_Optimizer.zero_grad |
2 | 0 | 0 |
meth |
BF16_Optimizer.state_dict |
1 | 0 | 0 |
meth |
BF16_Optimizer._restore_from_bit16_weights |
1 | 0 | 0 |
meth |
BF16_Optimizer.refresh_fp32_params |
1 | 0 | 0 |
meth |
BF16_Optimizer.load_state_dict |
7 | 0 | 0 |
meth |
BF16_Optimizer._load_legacy_checkpoint |
4 | 0 | 0 |
meth |
BF16_Optimizer._load_universal_checkpoint |
4 | 0 | 0 |
meth |
BF16_Optimizer._load_global_state |
2 | 0 | 0 |
meth |
BF16_Optimizer.accumulate_hp_grads_and_remove_lp |
4 | 0 | 0 |
meth |
BF16_Optimizer.create_grad_acc_hooks |
1 | 0 | 0 |
prop |
BF16_Optimizer.param_groups |
1 | 0 | 0 |
prop |
BF16_Optimizer.state |
1 | 0 | 0 |
attr |
BF16_Optimizer.timers |
1 | 0 | 0 |
attr |
BF16_Optimizer.optimizer |
1 | 0 | 0 |
attr |
BF16_Optimizer.param_names |
1 | 0 | 0 |
attr |
BF16_Optimizer.using_real_optimizer |
1 | 0 | 0 |
attr |
BF16_Optimizer.grad_acc_dtype |
1 | 0 | 0 |
attr |
BF16_Optimizer.custom_loss_scaler |
1 | 0 | 0 |
attr |
BF16_Optimizer.external_loss_scale |
1 | 0 | 0 |
attr |
BF16_Optimizer.torch_autocast_gradscaler |
1 | 0 | 0 |
attr |
BF16_Optimizer.immediate_grad_update |
1 | 0 | 0 |
attr |
BF16_Optimizer.clip_grad |
1 | 0 | 0 |
attr |
BF16_Optimizer.norm_type |
1 | 0 | 0 |
attr |
BF16_Optimizer.mpu |
1 | 0 | 0 |
attr |
BF16_Optimizer.allgather_bucket_size |
1 | 0 | 0 |
attr |
BF16_Optimizer.dp_process_group |
1 | 0 | 0 |
attr |
BF16_Optimizer.dp_rank |
1 | 0 | 0 |
attr |
BF16_Optimizer.has_moe_layers |
1 | 0 | 0 |
attr |
BF16_Optimizer.non_expert_gradients |
1 | 0 | 0 |
attr |
BF16_Optimizer.real_dp_process_group |
1 | 0 | 0 |
attr |
BF16_Optimizer.flatten |
1 | 0 | 0 |
attr |
BF16_Optimizer.unflatten |
1 | 0 | 0 |
attr |
BF16_Optimizer.nccl_start_alignment_factor |
1 | 0 | 0 |
attr |
BF16_Optimizer.bf16_groups |
1 | 0 | 0 |
attr |
BF16_Optimizer.bf16_groups_flat |
1 | 0 | 0 |
attr |
BF16_Optimizer.bf16_partitioned_groups |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_flat_partition |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_gradients |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_gradient_dict |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_gradients_flat |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_actual_gradients_flat |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_gradient_flat_partition |
1 | 0 | 0 |
attr |
BF16_Optimizer.fp32_groups_has_gradients |
1 | 0 | 0 |
attr |
BF16_Optimizer.group_paddings |
1 | 0 | 0 |
attr |
BF16_Optimizer.graph_harvesting |
1 | 0 | 0 |
func |
print_rank_0 |
4 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.checkpoint_engine (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CheckpointEngine.init |
2 | 0 | 0 |
meth |
CheckpointEngine.create |
2 | 1 | 0 |
meth |
CheckpointEngine.save |
3 | 1 | 0 |
meth |
CheckpointEngine.makedirs |
3 | 0 | 0 |
meth |
CheckpointEngine.load |
3 | 1 | 0 |
meth |
CheckpointEngine.commit |
2 | 1 | 0 |
meth |
CheckpointEngine.is_data_parallel_writer |
2 | 0 | 0 |
meth |
CheckpointEngine.is_decoupled |
1 | 0 | 0 |
meth |
CheckpointEngine.set_commit_info |
2 | 1 | 0 |
meth |
CheckpointEngine.get_commit_info |
1 | 0 | 0 |
meth |
CheckpointEngine.cleanup |
1 | 0 | 0 |
meth |
CheckpointEngine.preserves_storage_sharing |
1 | 0 | 0 |
attr |
CheckpointEngine.name |
1 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.datastates_checkpoint_engine (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DataStatesCheckpointEngine.init |
3 | 0 | 0 |
meth |
DataStatesCheckpointEngine.del |
1 | 0 | 0 |
meth |
DataStatesCheckpointEngine.create |
2 | 1 | 0 |
meth |
DataStatesCheckpointEngine.save |
3 | 1 | 0 |
meth |
DataStatesCheckpointEngine.load |
3 | 1 | 0 |
meth |
DataStatesCheckpointEngine.commit |
2 | 1 | 0 |
meth |
DataStatesCheckpointEngine.cleanup |
1 | 0 | 0 |
meth |
DataStatesCheckpointEngine.is_decoupled |
1 | 0 | 0 |
meth |
DataStatesCheckpointEngine.preserves_storage_sharing |
1 | 0 | 0 |
attr |
DataStatesCheckpointEngine.commit_info |
1 | 0 | 0 |
attr |
DataStatesCheckpointEngine.ckpt_engine |
1 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.decoupled_checkpoint_engine (41 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DecoupledCheckpointEngine.init |
4 | 0 | 0 |
meth |
DecoupledCheckpointEngine.del |
1 | 0 | 0 |
meth |
DecoupledCheckpointEngine._check_process_alive |
1 | 0 | 0 |
meth |
DecoupledCheckpointEngine._wait_for_event_with_timeout |
2 | 0 | 0 |
meth |
DecoupledCheckpointEngine.create |
2 | 1 | 0 |
meth |
DecoupledCheckpointEngine.load |
3 | 1 | 0 |
meth |
DecoupledCheckpointEngine.save |
3 | 1 | 0 |
meth |
DecoupledCheckpointEngine.commit |
2 | 1 | 0 |
meth |
DecoupledCheckpointEngine.get_commit_info |
1 | 0 | 0 |
meth |
DecoupledCheckpointEngine.is_decoupled |
1 | 0 | 0 |
meth |
DecoupledCheckpointEngine.cleanup |
1 | 0 | 0 |
meth |
DecoupledCheckpointEngine.is_data_parallel_writer |
2 | 0 | 0 |
attr |
DecoupledCheckpointEngine.name |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.dp_writer_config |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.commit_info |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.checkpoint_size |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.global_rank |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.optimize_dp_state |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.save_event |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.save_queue |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.ckpt_process |
1 | 0 | 0 |
attr |
DecoupledCheckpointEngine.local_rank |
1 | 0 | 0 |
func |
init_decoupled_checkpoint |
6 | 0 | 0 |
meth |
CheckpointSize.init |
1 | 0 | 0 |
meth |
CheckpointSize.gb_size |
1 | 0 | 0 |
meth |
CheckpointSize.set_pre_size |
2 | 0 | 0 |
meth |
CheckpointSize.set_post_size |
2 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.fast_checkpoint_engine (15 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FastCheckpointEngine.init |
4 | 0 | 0 |
meth |
FastCheckpointEngine.create |
2 | 1 | 0 |
meth |
FastCheckpointEngine.save |
3 | 1 | 0 |
meth |
FastCheckpointEngine.load |
3 | 1 | 0 |
meth |
FastCheckpointEngine.commit |
2 | 1 | 0 |
meth |
FastCheckpointEngine.is_data_parallel_writer |
2 | 0 | 0 |
attr |
FastCheckpointEngine.name |
1 | 0 | 0 |
attr |
FastCheckpointEngine.serialization_enabled |
1 | 0 | 0 |
attr |
FastCheckpointEngine.optimize_dp_state |
1 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.nebula_checkpoint_engine (13 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NebulaCheckpointEngine.init |
2 | 0 | 0 |
meth |
NebulaCheckpointEngine.create |
2 | 1 | 0 |
meth |
NebulaCheckpointEngine.save |
3 | 1 | 0 |
meth |
NebulaCheckpointEngine.load |
3 | 1 | 0 |
meth |
NebulaCheckpointEngine.commit |
2 | 1 | 0 |
attr |
NebulaCheckpointEngine.name |
1 | 0 | 0 |
attr |
NebulaCheckpointEngine.checkpoint |
1 | 0 | 0 |
attr |
NebulaCheckpointEngine.tag_flag |
1 | 0 | 0 |
attr |
NebulaCheckpointEngine.enable_nebula_load |
1 | 0 | 0 |
attr |
NebulaCheckpointEngine.nebula_load_path |
1 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.torch_checkpoint_engine (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TorchCheckpointEngine.init |
2 | 0 | 0 |
meth |
TorchCheckpointEngine.create |
2 | 1 | 0 |
meth |
TorchCheckpointEngine.save |
3 | 1 | 0 |
meth |
TorchCheckpointEngine.load |
3 | 1 | 0 |
meth |
TorchCheckpointEngine.commit |
2 | 1 | 0 |
attr |
TorchCheckpointEngine.name |
1 | 0 | 0 |
attr |
TorchCheckpointEngine.zipfile_serialization |
1 | 0 | 0 |
deepspeed.runtime.checkpoint_engine.utils (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
create_checkpoint_engine |
6 | 0 | 0 |
deepspeed.runtime.comm.coalesced_collectives (0 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
all_to_all_loco_quant_reduce |
4 | 4 | 1 |
deepspeed.runtime.comm.compressed (32 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CompressedBackend.init |
2 | 0 | 0 |
meth |
CompressedBackend.my_igather |
7 | 0 | 0 |
meth |
CompressedBackend.my_gather |
7 | 0 | 0 |
meth |
CompressedBackend.pack |
3 | 0 | 0 |
meth |
CompressedBackend.unpack |
4 | 0 | 0 |
meth |
CompressedBackend.compressed_allreduce |
5 | 1 | 0 |
attr |
CompressedBackend.size |
1 | 0 | 0 |
attr |
CompressedBackend.rank |
1 | 0 | 0 |
attr |
CompressedBackend.packer |
1 | 0 | 0 |
attr |
CompressedBackend.world_group |
1 | 0 | 0 |
attr |
CompressedBackend.mpu |
1 | 0 | 0 |
deepspeed.runtime.comm.hccl (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
HcclBackend.init |
2 | 0 | 0 |
meth |
HcclBackend.my_igather |
7 | 0 | 0 |
meth |
HcclBackend.my_gather |
7 | 0 | 0 |
meth |
HcclBackend.compressed_allreduce |
5 | 1 | 0 |
attr |
HcclBackend.size |
1 | 0 | 0 |
attr |
HcclBackend.rank |
1 | 0 | 0 |
attr |
HcclBackend.world_group |
1 | 0 | 0 |
attr |
HcclBackend.mpu |
1 | 0 | 0 |
deepspeed.runtime.comm.mpi (46 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MpiBackend.init |
2 | 0 | 0 |
meth |
MpiBackend.my_igather |
7 | 0 | 0 |
meth |
MpiBackend.gather_cuda |
8 | 0 | 0 |
meth |
MpiBackend.gather_host |
8 | 0 | 0 |
meth |
MpiBackend.allgather_cuda |
6 | 0 | 0 |
meth |
MpiBackend.allgather_host |
6 | 0 | 0 |
meth |
MpiBackend.compressed_allreduce |
5 | 1 | 0 |
attr |
MpiBackend.comm |
1 | 0 | 0 |
attr |
MpiBackend.rank |
1 | 0 | 0 |
attr |
MpiBackend.size |
1 | 0 | 0 |
attr |
MpiBackend.cuda_aware |
1 | 0 | 0 |
attr |
MpiBackend.compression_backend |
1 | 0 | 0 |
deepspeed.runtime.comm.nccl (26 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
NcclBackend.init |
2 | 0 | 0 |
meth |
NcclBackend.my_igather |
7 | 0 | 0 |
meth |
NcclBackend.my_gather |
7 | 0 | 0 |
meth |
NcclBackend.compressed_allreduce |
5 | 1 | 0 |
attr |
NcclBackend.rank |
1 | 0 | 0 |
attr |
NcclBackend.size |
1 | 0 | 0 |
attr |
NcclBackend.compression_backend |
1 | 0 | 0 |
attr |
NcclBackend.bool_not_supported |
1 | 0 | 0 |
attr |
NcclBackend.world_group |
1 | 0 | 0 |
attr |
NcclBackend.mpu |
1 | 0 | 0 |
deepspeed.runtime.compiler (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
dummy_decorator |
2 | 0 | 0 |
func |
compiled_autograd |
3 | 2 | 0 |
func |
is_compiling |
1 | 0 | 0 |
func |
compile |
1 | 0 | 0 |
func |
disable |
2 | 0 | 0 |
func |
is_compile_supported |
1 | 0 | 0 |
func |
enable |
2 | 0 | 0 |
deepspeed.runtime.compression.cupy (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CupyBackend.init |
1 | 0 | 0 |
meth |
CupyBackend.torch2cupy |
2 | 0 | 0 |
meth |
CupyBackend.cupy2torch |
2 | 0 | 0 |
meth |
CupyBackend.compress_by_chunk |
3 | 0 | 0 |
deepspeed.runtime.config (142 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_pld_params |
2 | 0 | 0 |
func |
get_disable_allgather |
2 | 0 | 0 |
func |
get_sparse_attention_type |
2 | 0 | 0 |
func |
get_scheduler_params |
2 | 0 | 0 |
func |
get_eigenvalue_verbose |
2 | 0 | 0 |
func |
get_sparse_attention_mode |
2 | 0 | 0 |
func |
get_checkpoint_tag_validation_mode |
2 | 0 | 0 |
func |
get_prescale_gradients |
2 | 0 | 0 |
func |
get_pld_enabled |
2 | 0 | 0 |
func |
get_gradient_clipping |
2 | 0 | 0 |
func |
get_optimizer_name |
2 | 0 | 0 |
func |
get_eigenvalue_stability |
2 | 0 | 0 |
func |
get_sparse_variable_config |
2 | 0 | 0 |
meth |
DeepSpeedConfigWriter.init |
2 | 0 | 0 |
meth |
DeepSpeedConfigWriter.add_config |
3 | 0 | 0 |
meth |
DeepSpeedConfigWriter.load_config |
2 | 0 | 0 |
meth |
DeepSpeedConfigWriter.write_config |
2 | 0 | 0 |
attr |
DeepSpeedConfigWriter.data |
1 | 0 | 0 |
meth |
DeepSpeedConfig.init |
4 | 1 | 0 |
meth |
DeepSpeedConfig._initialize_params |
2 | 0 | 0 |
meth |
DeepSpeedConfig._batch_assertion |
1 | 0 | 0 |
meth |
DeepSpeedConfig._set_batch_related_parameters |
1 | 0 | 0 |
meth |
DeepSpeedConfig._configure_train_batch_size |
1 | 0 | 0 |
meth |
DeepSpeedConfig._do_sanity_check |
1 | 0 | 0 |
meth |
DeepSpeedConfig.print_user_config |
1 | 0 | 0 |
meth |
DeepSpeedConfig.print |
2 | 0 | 0 |
meth |
DeepSpeedConfig._do_error_check |
1 | 0 | 0 |
meth |
DeepSpeedConfig._do_warning_check |
1 | 0 | 0 |
attr |
DeepSpeedConfig.elasticity_enabled |
1 | 0 | 0 |
attr |
DeepSpeedConfig.global_rank |
1 | 0 | 0 |
attr |
DeepSpeedConfig.elastic_model_parallel_size |
1 | 0 | 0 |
attr |
DeepSpeedConfig.num_gpus_per_node |
1 | 0 | 0 |
attr |
DeepSpeedConfig.world_size |
1 | 0 | 0 |
func |
get_train_micro_batch_size_per_gpu |
2 | 0 | 0 |
func |
get_train_batch_size |
2 | 0 | 0 |
func |
get_steps_per_print |
2 | 0 | 0 |
func |
get_dump_state |
2 | 0 | 0 |
func |
get_sparse_fixed_config |
2 | 0 | 0 |
func |
get_amp_params |
2 | 0 | 0 |
func |
get_torch_autocast_enabled |
2 | 0 | 0 |
func |
get_hybrid_engine_config |
2 | 0 | 0 |
func |
get_scheduler_name |
2 | 0 | 0 |
func |
get_sparse_bigbird_config |
2 | 0 | 0 |
func |
get_optimizer_params |
2 | 0 | 0 |
func |
get_pipeline_config |
2 | 0 | 0 |
func |
get_checkpoint_params |
2 | 0 | 0 |
func |
get_wall_clock_breakdown |
2 | 0 | 0 |
func |
get_zero_force_ds_cpu_optimizer |
2 | 0 | 0 |
func |
get_gradient_accumulation_steps |
2 | 0 | 0 |
func |
get_sparse_gradients_enabled |
2 | 0 | 0 |
func |
get_zero_allow_untested_optimizer |
2 | 0 | 0 |
func |
get_eigenvalue_layer_name |
2 | 0 | 0 |
func |
get_eigenvalue_max_iter |
2 | 0 | 0 |
meth |
DtypeEnum.new |
2 | 0 | 0 |
meth |
DtypeEnum.repr |
1 | 0 | 0 |
func |
get_amp_enabled |
2 | 0 | 0 |
func |
get_memory_breakdown |
2 | 0 | 0 |
func |
get_checkpoint_parallel_write_pipeline |
2 | 0 | 0 |
func |
get_optimizer_gradient_clipping |
2 | 0 | 0 |
func |
get_eigenvalue_enabled |
2 | 0 | 0 |
func |
get_gradient_predivide_factor |
2 | 0 | 0 |
func |
get_dataloader_drop_last |
2 | 0 | 0 |
func |
get_torch_autocast_dtype |
2 | 0 | 0 |
func |
get_eigenvalue_config |
2 | 0 | 0 |
func |
get_lower_precision_safe_modules |
2 | 0 | 0 |
func |
get_eigenvalue_tol |
2 | 0 | 0 |
func |
get_eigenvalue_layer_num |
2 | 0 | 0 |
func |
get_graph_harvesting |
2 | 0 | 0 |
func |
get_optimizer_legacy_fusion |
2 | 0 | 0 |
func |
get_sparse_dense_config |
2 | 0 | 0 |
func |
get_communication_data_type |
4 | 0 | 0 |
func |
get_expert_data_topo_config |
2 | 0 | 0 |
func |
get_eigenvalue_gas_boundary_resolution |
2 | 0 | 0 |
func |
get_data_types_params |
2 | 0 | 0 |
func |
get_sparse_attention |
2 | 0 | 0 |
func |
get_sparse_bslongformer_config |
2 | 0 | 0 |
deepspeed.runtime.config_utils (34 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_scalar_param |
4 | 0 | 0 |
func |
get_config_default |
3 | 0 | 0 |
func |
get_list_param |
4 | 0 | 0 |
meth |
ScientificNotationEncoder.iterencode |
4 | 0 | 0 |
meth |
DeepSpeedConfigObject.repr |
1 | 0 | 0 |
meth |
DeepSpeedConfigObject.repr |
1 | 0 | 0 |
meth |
pp_int.new |
3 | 0 | 0 |
meth |
pp_int.repr |
1 | 0 | 0 |
func |
get_dict_param |
4 | 0 | 0 |
meth |
DeepSpeedConfigModel.init |
3 | 0 | 0 |
meth |
DeepSpeedConfigModel._process_deprecated_field |
2 | 0 | 0 |
meth |
DeepSpeedConfigModel._deprecated_fields_check |
1 | 0 | 0 |
attr |
DeepSpeedConfigModel.model_config |
1 | 0 | 0 |
func |
dict_raise_error_on_duplicate_keys |
2 | 0 | 0 |
deepspeed.runtime.data_pipeline.config (38 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_data_sampling |
2 | 0 | 0 |
func |
get_data_sampling_enabled |
2 | 0 | 0 |
func |
get_random_ltd_enabled |
2 | 0 | 0 |
func |
get_dynamic_batching |
2 | 0 | 0 |
func |
get_data_routing_enabled |
2 | 0 | 0 |
func |
get_curriculum_learning_enabled |
2 | 0 | 0 |
func |
get_data_sampling_num_epochs |
2 | 0 | 0 |
func |
get_data_efficiency_enabled |
2 | 0 | 0 |
func |
get_random_ltd |
2 | 0 | 0 |
func |
get_data_routing |
2 | 0 | 0 |
func |
get_curriculum_learning_params |
2 | 0 | 0 |
func |
get_curriculum_learning |
2 | 0 | 0 |
func |
get_random_ltd_params |
2 | 0 | 0 |
func |
get_data_efficiency_config |
2 | 0 | 0 |
func |
get_curriculum_enabled_legacy |
2 | 0 | 0 |
func |
get_data_efficiency_seed |
2 | 0 | 0 |
func |
get_curriculum_params_legacy |
2 | 0 | 0 |
func |
get_data_sampling_num_workers |
2 | 0 | 0 |
func |
get_data_sampling_pin_memory |
2 | 0 | 0 |
deepspeed.runtime.data_pipeline.curriculum_scheduler (22 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CurriculumScheduler.init |
2 | 0 | 0 |
meth |
CurriculumScheduler.get_current_difficulty |
1 | 0 | 0 |
meth |
CurriculumScheduler.set_current_difficulty |
2 | 0 | 0 |
meth |
CurriculumScheduler.set_custom_get_difficulty |
2 | 0 | 0 |
meth |
CurriculumScheduler.get_state |
1 | 0 | 0 |
meth |
CurriculumScheduler.set_state |
2 | 0 | 0 |
meth |
CurriculumScheduler.__fixed_discrete_get_difficulty |
2 | 0 | 0 |
meth |
CurriculumScheduler.__fixed_root_get_difficulty |
3 | 0 | 0 |
meth |
CurriculumScheduler.get_difficulty |
2 | 0 | 0 |
meth |
CurriculumScheduler.update_difficulty |
2 | 0 | 0 |
attr |
CurriculumScheduler.state |
1 | 0 | 0 |
attr |
CurriculumScheduler.first_step |
1 | 0 | 0 |
attr |
CurriculumScheduler.custom_get_difficulty |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_routing.basic_layer (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RandomLayerTokenDrop.init |
2 | 1 | 0 |
meth |
RandomLayerTokenDrop.init_config |
4 | 0 | 0 |
meth |
RandomLayerTokenDrop.get_bsh |
2 | 0 | 0 |
meth |
RandomLayerTokenDrop.get_sbh |
2 | 0 | 0 |
meth |
RandomLayerTokenDrop.forward |
3 | 1 | 0 |
attr |
RandomLayerTokenDrop.random_ltd_layer |
1 | 0 | 0 |
attr |
RandomLayerTokenDrop.reserved_length |
1 | 0 | 0 |
attr |
RandomLayerTokenDrop.random_ltd_scheduler |
1 | 0 | 0 |
attr |
RandomLayerTokenDrop.max_length |
1 | 0 | 0 |
attr |
RandomLayerTokenDrop.curr_seq |
1 | 0 | 0 |
attr |
RandomLayerTokenDrop.batch_first |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_routing.helper (7 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
convert_to_random_ltd |
3 | 0 | 0 |
func |
remove_random_ltd_state_dict |
2 | 0 | 0 |
func |
save_without_random_ltd |
2 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_routing.scheduler (28 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RandomLTDScheduler.init |
2 | 0 | 0 |
meth |
RandomLTDScheduler.get_total_layer_tokens |
2 | 0 | 0 |
meth |
RandomLTDScheduler.reset_to_init |
1 | 0 | 0 |
meth |
RandomLTDScheduler.get_current_seq |
1 | 0 | 0 |
meth |
RandomLTDScheduler.set_current_seq |
2 | 0 | 0 |
meth |
RandomLTDScheduler.get_random_ltd_layer_num |
1 | 0 | 0 |
meth |
RandomLTDScheduler.get_state |
1 | 0 | 0 |
meth |
RandomLTDScheduler.set_state |
2 | 0 | 0 |
meth |
RandomLTDScheduler.update_seq |
2 | 0 | 0 |
meth |
RandomLTDScheduler.state_dict |
1 | 0 | 0 |
meth |
RandomLTDScheduler.load_state_dict |
2 | 0 | 0 |
attr |
RandomLTDScheduler.model_layer_num |
1 | 0 | 0 |
attr |
RandomLTDScheduler.random_ltd_layer_num |
1 | 0 | 0 |
attr |
RandomLTDScheduler.config_schedule |
1 | 0 | 0 |
attr |
RandomLTDScheduler.global_batch_size |
1 | 0 | 0 |
meth |
BaseScheduler.init |
1 | 0 | 0 |
meth |
BaseScheduler.__fixed_root_get_value |
3 | 0 | 0 |
meth |
BaseScheduler.get_value |
2 | 0 | 0 |
attr |
BaseScheduler.state |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_routing.utils (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
bsh_decoder_scatter |
4 | 0 | 0 |
func |
bsh_decoder_gather |
4 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_sampling.data_analyzer (157 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DistributedDataAnalyzer.init |
14 | 1 | 0 |
meth |
DistributedDataAnalyzer.run_map_helper |
3 | 0 | 0 |
meth |
DistributedDataAnalyzer.run_map_reduce |
1 | 0 | 0 |
meth |
DistributedDataAnalyzer.file_write_ordered |
4 | 0 | 0 |
attr |
DistributedDataAnalyzer.dataset |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.batch_size |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.metric_names |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.metric_functions |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.metric_types |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.save_path |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.collate_fn |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.device |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.sample_indices |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.num_threads |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.worker_id |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.comm_group |
1 | 0 | 0 |
attr |
DistributedDataAnalyzer.num_workers |
1 | 0 | 0 |
func |
test_compare_both_data_analyzers |
2 | 0 | 0 |
meth |
DataAnalyzer.init |
19 | 0 | 0 |
meth |
DataAnalyzer.init_metric_results |
7 | 0 | 0 |
meth |
DataAnalyzer.update_metric_results |
7 | 0 | 0 |
meth |
DataAnalyzer.finalize_metric_results |
4 | 0 | 0 |
meth |
DataAnalyzer.run_map_helper |
2 | 0 | 0 |
meth |
DataAnalyzer.run_map |
1 | 0 | 0 |
meth |
DataAnalyzer.get_metric_value_percentiles |
4 | 0 | 0 |
meth |
DataAnalyzer.merge_gather_map_stats |
8 | 0 | 0 |
meth |
DataAnalyzer.merge_sample_to_metric |
6 | 0 | 0 |
meth |
DataAnalyzer.merge_metric_to_sample |
9 | 0 | 0 |
meth |
DataAnalyzer.merge_map_results |
8 | 0 | 0 |
meth |
DataAnalyzer.output_index_to_sample_percentile |
7 | 0 | 0 |
meth |
DataAnalyzer.run_reduce |
1 | 0 | 0 |
meth |
DataAnalyzer.run_map_reduce |
2 | 0 | 0 |
attr |
DataAnalyzer.dataset |
1 | 0 | 0 |
attr |
DataAnalyzer.num_workers |
1 | 0 | 0 |
attr |
DataAnalyzer.worker_id |
1 | 0 | 0 |
attr |
DataAnalyzer.num_threads |
1 | 0 | 0 |
attr |
DataAnalyzer.num_threads_reduce |
1 | 0 | 0 |
attr |
DataAnalyzer.specific_threads |
1 | 0 | 0 |
attr |
DataAnalyzer.batch_size |
1 | 0 | 0 |
attr |
DataAnalyzer.metric_names |
1 | 0 | 0 |
attr |
DataAnalyzer.metric_functions |
1 | 0 | 0 |
attr |
DataAnalyzer.metric_types |
1 | 0 | 0 |
attr |
DataAnalyzer.metric_dtypes |
1 | 0 | 0 |
attr |
DataAnalyzer.save_path |
1 | 0 | 0 |
attr |
DataAnalyzer.collate_fn |
1 | 0 | 0 |
attr |
DataAnalyzer.custom_map_init |
1 | 0 | 0 |
attr |
DataAnalyzer.custom_map_update |
1 | 0 | 0 |
attr |
DataAnalyzer.custom_map_finalize |
1 | 0 | 0 |
attr |
DataAnalyzer.custom_reduce |
1 | 0 | 0 |
attr |
DataAnalyzer.sample_indices |
1 | 0 | 0 |
meth |
Dist.min_max |
3 | 0 | 0 |
meth |
Dist.gather_v |
6 | 0 | 0 |
meth |
Dist.sample_sort |
5 | 0 | 0 |
meth |
TestDataset.init |
2 | 0 | 0 |
attr |
TestDataset.values |
1 | 0 | 0 |
attr |
TestDataset.size |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_sampling.data_sampler (64 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedDataSampler.init |
10 | 0 | 0 |
meth |
DeepSpeedDataSampler.len |
1 | 0 | 0 |
meth |
DeepSpeedDataSampler.set_custom_curriculum_learning_schedule |
2 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_start_end_idx |
2 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_sample_based_on_metric_value |
4 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_sample_based_on_metric_percentile |
4 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_new_cluster |
2 | 0 | 0 |
meth |
DeepSpeedDataSampler.sample_from_clusters |
1 | 0 | 0 |
meth |
DeepSpeedDataSampler.reshuffle_clusters |
2 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_sample_from_cluster |
3 | 0 | 0 |
meth |
DeepSpeedDataSampler.get_next_global_batch |
1 | 0 | 0 |
meth |
DeepSpeedDataSampler.iter |
1 | 0 | 0 |
meth |
DeepSpeedDataSampler.state_dict |
1 | 0 | 0 |
meth |
DeepSpeedDataSampler.load_state_dict |
2 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_efficiency_config |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.one_epoch_total_samples |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.index_dtype |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.total_samples |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.micro_batch_size |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_parallel_rank |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_parallel_group |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.micro_batch_times_data_parallel_size |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.gradient_accumulation_steps |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.global_batch_size |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.global_rank |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.drop_last |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.np_rng |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.state |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.batch |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.consumed_samples |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.curriculum_step |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.current_difficulties |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_cluster_paths |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_cluster_current_position |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.curriculum_schedulers |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.curriculum_index_to_sample |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.curriculum_index_to_metric |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.difficulty_type |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.clustering_type |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_1epoch_size |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_clusters |
1 | 0 | 0 |
attr |
DeepSpeedDataSampler.data_cluster_sizes |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_sampling.indexed_dataset (137 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
write_longs |
3 | 0 | 0 |
func |
dataset_exists |
3 | 0 | 0 |
meth |
IndexedCachedDataset.init |
2 | 0 | 0 |
meth |
IndexedCachedDataset.prefetch |
2 | 0 | 0 |
meth |
IndexedCachedDataset.getitem |
2 | 0 | 0 |
prop |
IndexedCachedDataset.supports_prefetch |
1 | 0 | 0 |
attr |
IndexedCachedDataset.cache |
1 | 0 | 0 |
attr |
IndexedCachedDataset.cache_index |
1 | 0 | 0 |
func |
code |
2 | 0 | 0 |
meth |
MMapIndexedDataset.init |
3 | 0 | 0 |
meth |
MMapIndexedDataset.getstate |
1 | 0 | 0 |
meth |
MMapIndexedDataset.setstate |
2 | 0 | 0 |
meth |
MMapIndexedDataset.do_init |
3 | 0 | 0 |
meth |
MMapIndexedDataset.__del__ |
1 | 0 | 0 |
meth |
MMapIndexedDataset.__len__ |
1 | 0 | 0 |
meth |
MMapIndexedDataset.__getitem__ |
2 | 0 | 0 |
meth |
MMapIndexedDataset.get |
4 | 0 | 0 |
meth |
MMapIndexedDataset.size |
2 | 0 | 0 |
meth |
MMapIndexedDataset.get_doc_idx |
1 | 0 | 0 |
meth |
MMapIndexedDataset.set_doc_idx |
2 | 0 | 0 |
meth |
MMapIndexedDataset.exists |
2 | 0 | 0 |
prop |
MMapIndexedDataset.sizes |
1 | 0 | 0 |
prop |
MMapIndexedDataset.doc_idx |
1 | 0 | 0 |
prop |
MMapIndexedDataset.supports_prefetch |
1 | 0 | 0 |
prop |
MMapIndexedDataset.dtype |
1 | 0 | 0 |
meth |
Index.writer |
3 | 0 | 0 |
meth |
Index.__init__ |
3 | 0 | 0 |
meth |
Index.__del__ |
1 | 0 | 0 |
meth |
Index.__getitem__ |
2 | 0 | 0 |
meth |
Index.__len__ |
1 | 0 | 0 |
prop |
Index.dtype |
1 | 0 | 0 |
prop |
Index.sizes |
1 | 0 | 0 |
prop |
Index.doc_idx |
1 | 0 | 0 |
func |
make_dataset |
4 | 0 | 0 |
meth |
IndexedDataset.__init__ |
2 | 0 | 0 |
meth |
IndexedDataset.read_index |
2 | 0 | 0 |
meth |
IndexedDataset.read_data |
2 | 0 | 0 |
meth |
IndexedDataset.check_index |
2 | 0 | 0 |
meth |
IndexedDataset.__del__ |
1 | 0 | 0 |
meth |
IndexedDataset.__getitem__ |
2 | 0 | 0 |
meth |
IndexedDataset.__len__ |
1 | 0 | 0 |
meth |
IndexedDataset.num_tokens |
2 | 0 | 0 |
meth |
IndexedDataset.size |
2 | 0 | 0 |
meth |
IndexedDataset.exists |
2 | 0 | 0 |
prop |
IndexedDataset.supports_prefetch |
1 | 0 | 0 |
attr |
IndexedDataset.path |
1 | 0 | 0 |
attr |
IndexedDataset.data_file |
1 | 0 | 0 |
func |
index_file_path |
2 | 0 | 0 |
func |
create_doc_idx |
2 | 0 | 0 |
func |
exscan_from_cumsum |
2 | 0 | 0 |
func |
data_file_path |
2 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.init |
3 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.add_item |
2 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.add_items |
2 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.add_item_numpy |
2 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.end_document |
1 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.merge_file_ |
2 | 0 | 0 |
meth |
MMapIndexedDatasetBuilder.finalize |
2 | 0 | 0 |
func |
get_pointers_with_total |
4 | 0 | 0 |
func |
infer_dataset_impl |
2 | 0 | 0 |
attr |
valid_dtypes |
1 | 0 | 0 |
func |
read_longs |
3 | 0 | 0 |
func |
make_builder |
4 | 0 | 0 |
meth |
IndexedDatasetBuilder.init |
3 | 0 | 0 |
meth |
IndexedDatasetBuilder.add_item |
2 | 0 | 0 |
meth |
IndexedDatasetBuilder.end_document |
1 | 0 | 0 |
meth |
IndexedDatasetBuilder.merge_file_ |
2 | 0 | 0 |
meth |
IndexedDatasetBuilder.finalize |
2 | 0 | 0 |
attr |
IndexedDatasetBuilder.out_file |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.dtype |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.data_offsets |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.dim_offsets |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.sizes |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.element_size |
1 | 0 | 0 |
attr |
IndexedDatasetBuilder.doc_idx |
1 | 0 | 0 |
func |
get_available_dataset_impl |
1 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_sampling.utils (18 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
close_mmap_dataset_builder |
3 | 0 | 0 |
func |
create_mmap_dataset_builder |
3 | 0 | 0 |
func |
split_dataset |
5 | 0 | 0 |
func |
split_index |
4 | 0 | 0 |
func |
find_fit_int_dtype |
3 | 0 | 0 |
deepspeed.runtime.data_pipeline.data_sampling.variable_batch_size_and_lr (89 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
scale_lr |
5 | 0 | 0 |
func |
lr_scheduler_for_variable_batch_size |
7 | 0 | 0 |
func |
dataloader_for_variable_batch_size |
12 | 0 | 0 |
func |
get_dataloader_and_lr_scheduler_for_variable_batch_size_deepspeed |
8 | 0 | 0 |
meth |
VariableBatchSizeLR.init |
8 | 0 | 0 |
meth |
VariableBatchSizeLR.state_dict |
1 | 0 | 0 |
meth |
VariableBatchSizeLR.load_state_dict |
2 | 0 | 0 |
meth |
VariableBatchSizeLR.get_last_lr |
1 | 0 | 0 |
meth |
VariableBatchSizeLR.get_lr |
1 | 0 | 0 |
meth |
VariableBatchSizeLR.step |
2 | 0 | 0 |
prop |
VariableBatchSizeLR.optimizer |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.batch_sizes |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.base_batch_size |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.lr_scaling_method |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.dataloader |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.base_lr_scheduler |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.base_lrs |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.last_epoch |
1 | 0 | 0 |
attr |
VariableBatchSizeLR.verbose |
1 | 0 | 0 |
func |
batch_by_seqlens |
11 | 0 | 0 |
func |
get_dataloader_and_lr_scheduler_for_variable_batch_size |
22 | 0 | 0 |
deepspeed.runtime.dataloader (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
RepeatingLoader.init |
2 | 0 | 0 |
meth |
RepeatingLoader.iter |
1 | 0 | 0 |
meth |
RepeatingLoader.next |
1 | 0 | 0 |
attr |
RepeatingLoader.loader |
1 | 0 | 0 |
attr |
RepeatingLoader.data_iter |
1 | 0 | 0 |
meth |
DeepSpeedDataLoader.init |
13 | 0 | 0 |
meth |
DeepSpeedDataLoader.iter |
1 | 0 | 0 |
meth |
DeepSpeedDataLoader.len |
1 | 0 | 0 |
meth |
DeepSpeedDataLoader.next |
1 | 0 | 0 |
meth |
DeepSpeedDataLoader._create_dataloader |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.deepspeed_dataloader_config |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.tput_timer |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.batch_size |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.curriculum_learning_enabled |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.num_local_io_workers |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.data_sampler |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.dataset |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.collate_fn |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.device_count |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.pin_memory |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.data |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.dataloader_drop_last |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.post_process_func |
1 | 0 | 0 |
attr |
DeepSpeedDataLoader.len |
1 | 0 | 0 |
deepspeed.runtime.domino.async_linear (30 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DominoAsyncColumnParallelLinearImpl.forward |
7 | 0 | 0 |
meth |
DominoAsyncColumnParallelLinearImpl.backward |
3 | 0 | 0 |
meth |
DominoAsyncColumnParallelLinear.init |
8 | 1 | 0 |
meth |
DominoAsyncColumnParallelLinear.forward |
4 | 1 | 0 |
attr |
DominoAsyncColumnParallelLinear.skip_bias_add |
1 | 0 | 0 |
attr |
DominoAsyncColumnParallelLinear.weight |
1 | 0 | 0 |
attr |
DominoAsyncColumnParallelLinear.bias |
1 | 0 | 0 |
meth |
RowParallelLinearNoComm.init |
8 | 6 | 0 |
meth |
RowParallelLinearNoComm.forward |
2 | 0 | 0 |
attr |
RowParallelLinearNoComm.skip_bias_add |
1 | 0 | 0 |
attr |
RowParallelLinearNoComm.weight |
1 | 0 | 0 |
attr |
RowParallelLinearNoComm.bias |
1 | 0 | 0 |
deepspeed.runtime.domino.transformer (119 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CoreAttention.init |
4 | 0 | 0 |
meth |
CoreAttention.forward |
5 | 0 | 0 |
attr |
CoreAttention.attn_mask_type |
1 | 0 | 0 |
attr |
CoreAttention.hidden_size_per_partition |
1 | 0 | 0 |
attr |
CoreAttention.attention_dropout_rate |
1 | 0 | 0 |
meth |
ShardedAttention.init |
7 | 0 | 0 |
meth |
ShardedAttention.forward |
5 | 0 | 0 |
meth |
ShardedAttention.domino_core_attention_forward |
4 | 0 | 0 |
attr |
ShardedAttention.layer_number |
1 | 0 | 0 |
attr |
ShardedAttention.attention_type |
1 | 0 | 0 |
attr |
ShardedAttention.attn_mask_type |
1 | 0 | 0 |
attr |
ShardedAttention.params_dtype |
1 | 0 | 0 |
attr |
ShardedAttention.apply_rotary_pos_emb |
1 | 0 | 0 |
attr |
ShardedAttention.hidden_size_per_attention_head |
1 | 0 | 0 |
attr |
ShardedAttention.num_attention_heads_per_partition |
1 | 0 | 0 |
attr |
ShardedAttention.query_key_value |
1 | 0 | 0 |
attr |
ShardedAttention.core_attention |
1 | 0 | 0 |
attr |
ShardedAttention.dense |
1 | 0 | 0 |
meth |
DominoTransformer.init |
11 | 0 | 0 |
meth |
DominoTransformer.forward |
4 | 0 | 0 |
meth |
DominoTransformer.inter_layer_overlap_forward |
4 | 0 | 0 |
meth |
DominoTransformer.intra_layer_overlap_forward |
4 | 0 | 0 |
attr |
DominoTransformer.layer_type |
1 | 0 | 0 |
attr |
DominoTransformer.model_type |
1 | 0 | 0 |
attr |
DominoTransformer.post_layer_norm |
1 | 0 | 0 |
attr |
DominoTransformer.post_process |
1 | 0 | 0 |
attr |
DominoTransformer.input_tensor |
1 | 0 | 0 |
attr |
DominoTransformer.drop_path_rate |
1 | 0 | 0 |
attr |
DominoTransformer.TP_group |
1 | 0 | 0 |
attr |
DominoTransformer.num_layers |
1 | 0 | 0 |
attr |
DominoTransformer.drop_path_rates |
1 | 0 | 0 |
attr |
DominoTransformer.layers |
1 | 0 | 0 |
attr |
DominoTransformer.final_layernorm |
1 | 0 | 0 |
meth |
NoOper.symbolic |
5 | 0 | 0 |
meth |
NoOper.forward |
5 | 0 | 0 |
meth |
NoOper.backward |
3 | 0 | 0 |
meth |
DominoTransformerLayer.init |
8 | 0 | 0 |
meth |
DominoTransformerLayer.forward |
4 | 0 | 0 |
attr |
DominoTransformerLayer.layer_number |
1 | 0 | 0 |
attr |
DominoTransformerLayer.layer_type |
1 | 0 | 0 |
attr |
DominoTransformerLayer.apply_residual_connection_post_layernorm |
1 | 0 | 0 |
attr |
DominoTransformerLayer.llama_model |
1 | 0 | 0 |
attr |
DominoTransformerLayer.input_layernorm |
1 | 0 | 0 |
attr |
DominoTransformerLayer.self_attention |
1 | 0 | 0 |
attr |
DominoTransformerLayer.hidden_dropout |
1 | 0 | 0 |
attr |
DominoTransformerLayer.post_attention_layernorm |
1 | 0 | 0 |
attr |
DominoTransformerLayer.output_size_c |
1 | 0 | 0 |
attr |
DominoTransformerLayer.input_size_c |
1 | 0 | 0 |
attr |
DominoTransformerLayer.input_size_r |
1 | 0 | 0 |
attr |
DominoTransformerLayer.output_size_r |
1 | 0 | 0 |
attr |
DominoTransformerLayer.TP_group |
1 | 0 | 0 |
attr |
DominoTransformerLayer.output_size_per_partition |
1 | 0 | 0 |
attr |
DominoTransformerLayer.input_size_per_partition |
1 | 0 | 0 |
attr |
DominoTransformerLayer.linear_fc1 |
1 | 0 | 0 |
attr |
DominoTransformerLayer.mlp_activation_func |
1 | 0 | 0 |
attr |
DominoTransformerLayer.linear_fc2 |
1 | 0 | 0 |
attr |
DominoTransformerLayer.bias_dropout_add_func |
1 | 0 | 0 |
meth |
DominoModule.init |
1 | 0 | 0 |
meth |
bias_dropout_add.init |
2 | 1 | 0 |
attr |
bias_dropout_add.dropout |
1 | 0 | 0 |
deepspeed.runtime.eigenvalue (30 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Eigenvalue.init |
8 | 0 | 0 |
meth |
Eigenvalue.nan_to_num |
2 | 0 | 0 |
meth |
Eigenvalue.normalize |
2 | 0 | 0 |
meth |
Eigenvalue.inner_product |
3 | 0 | 0 |
meth |
Eigenvalue.get_layers |
2 | 0 | 0 |
meth |
Eigenvalue.compute_eigenvalue |
4 | 0 | 0 |
meth |
Eigenvalue.post_process |
2 | 0 | 0 |
attr |
Eigenvalue.verbose |
1 | 0 | 0 |
attr |
Eigenvalue.max_iter |
1 | 0 | 0 |
attr |
Eigenvalue.tol |
1 | 0 | 0 |
attr |
Eigenvalue.stability |
1 | 0 | 0 |
attr |
Eigenvalue.gas_boundary_resolution |
1 | 0 | 0 |
attr |
Eigenvalue.layer_name |
1 | 0 | 0 |
attr |
Eigenvalue.layer_num |
1 | 0 | 0 |
deepspeed.runtime.engine (545 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
split_half_float_double_sparse |
2 | 0 | 0 |
meth |
DeepSpeedEngine.init |
14 | 0 | 0 |
meth |
DeepSpeedEngine._optimized_linear_offload_setup |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_tensor_parallel |
3 | 0 | 0 |
meth |
DeepSpeedEngine._configure_tensor_parallel_states |
2 | 0 | 0 |
meth |
DeepSpeedEngine._apply_autotp_partitioning |
3 | 0 | 0 |
meth |
DeepSpeedEngine.del |
1 | 0 | 0 |
meth |
DeepSpeedEngine.destroy |
1 | 0 | 0 |
meth |
DeepSpeedEngine._get_model_parameters |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_batch_info |
1 | 0 | 0 |
meth |
DeepSpeedEngine.set_train_batch_size |
2 | 0 | 0 |
meth |
DeepSpeedEngine.set_train_micro_batch_size |
2 | 0 | 0 |
meth |
DeepSpeedEngine.set_data_post_process_func |
2 | 0 | 0 |
meth |
DeepSpeedEngine.set_custom_curriculum_learning_schedule |
2 | 0 | 0 |
meth |
DeepSpeedEngine.getattr |
2 | 0 | 0 |
meth |
DeepSpeedEngine.checkpoint_serialization_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.checkpoint_writer_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.checkpoint_tag_validation_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.checkpoint_tag_validation_fail |
1 | 0 | 0 |
meth |
DeepSpeedEngine.elasticity_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.is_elastic_model_parallel_supported |
1 | 0 | 0 |
meth |
DeepSpeedEngine.pld_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.pld_params |
1 | 0 | 0 |
meth |
DeepSpeedEngine.pld_theta |
1 | 0 | 0 |
meth |
DeepSpeedEngine.pld_gamma |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_verbose |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_max_iter |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_tol |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_stability |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_gas_boundary_resolution |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_layer_name |
1 | 0 | 0 |
meth |
DeepSpeedEngine.eigenvalue_layer_num |
1 | 0 | 0 |
meth |
DeepSpeedEngine.curriculum_enabled_legacy |
1 | 0 | 0 |
meth |
DeepSpeedEngine.curriculum_params_legacy |
1 | 0 | 0 |
meth |
DeepSpeedEngine.data_efficiency_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.data_efficiency_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.data_sampling_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.data_sampling_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.curriculum_learning_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.curriculum_learning_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.random_ltd_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.random_ltd_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.random_ltd_initialize |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_data_parallel_rank |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_tensor_parallel_rank |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_model_parallel_rank |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_sequence_parallel_group |
1 | 0 | 0 |
meth |
DeepSpeedEngine.wall_clock_breakdown |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_recompute_fwd_factor |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_profile_step |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_module_depth |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_top_modules |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_detailed |
1 | 0 | 0 |
meth |
DeepSpeedEngine.flops_profiler_output_file |
1 | 0 | 0 |
meth |
DeepSpeedEngine.memory_breakdown |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_start_profile_step |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_end_profile_step |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_metric_path |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_model_info_path |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_metric |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotuning_profile_model_info |
1 | 0 | 0 |
meth |
DeepSpeedEngine.sparse_gradients_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.train_batch_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.train_micro_batch_size_per_gpu |
1 | 0 | 0 |
meth |
DeepSpeedEngine.optimizer_name |
1 | 0 | 0 |
meth |
DeepSpeedEngine.optimizer_params |
1 | 0 | 0 |
meth |
DeepSpeedEngine.optimizer_legacy_fusion |
1 | 0 | 0 |
meth |
DeepSpeedEngine.scheduler_name |
1 | 0 | 0 |
meth |
DeepSpeedEngine.scheduler_params |
1 | 0 | 0 |
meth |
DeepSpeedEngine.quantize_training |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_optimization |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_allow_untested_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_force_ds_cpu_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_reduce_scatter |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_overlap_comm |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_offload_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_offload_param |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_use_cpu_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_cpu_offload |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_partial_offload |
1 | 0 | 0 |
meth |
DeepSpeedEngine.super_offload |
1 | 0 | 0 |
meth |
DeepSpeedEngine.cpuadam_cores_perc |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_sub_group_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_optimization_stage |
1 | 0 | 0 |
meth |
DeepSpeedEngine.mics_shard_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_reduce_bucket_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_multi_rank_bucket_allreduce |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_allgather_bucket_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_optimization_partition_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_optimization_partition_weights |
1 | 0 | 0 |
meth |
DeepSpeedEngine.is_first_weights_partition_group |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_contiguous_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_load_from_fp32_weights |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_elastic_checkpoint |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_nvme_offload_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_max_live_parameters |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_max_reuse_distance |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_prefetch_bucket_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_module_granularity_threshold |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_param_persistence_threshold |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_model_persistence_threshold |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_gather_16bit_weights_on_model_save |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_grad_hooks |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_legacy_stage1 |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_ignore_unused_parameters |
1 | 0 | 0 |
meth |
DeepSpeedEngine.tensor_parallel_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.autotp_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.graph_harvesting |
1 | 0 | 0 |
meth |
DeepSpeedEngine.fp16_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.bfloat16_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.fp16_master_weights_and_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.bf16_master_weights_and_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.bf16_optimizer_states |
1 | 0 | 0 |
meth |
DeepSpeedEngine.amp_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.amp_params |
1 | 0 | 0 |
meth |
DeepSpeedEngine.fp16_auto_cast |
1 | 0 | 0 |
meth |
DeepSpeedEngine.loss_scale |
1 | 0 | 0 |
meth |
DeepSpeedEngine.gradient_accumulation_steps |
1 | 0 | 0 |
meth |
DeepSpeedEngine.use_node_local_storage |
1 | 0 | 0 |
meth |
DeepSpeedEngine.load_universal_checkpoint |
1 | 0 | 0 |
meth |
DeepSpeedEngine.postscale_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.gradient_predivide_factor |
1 | 0 | 0 |
meth |
DeepSpeedEngine.steps_per_print |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_allgather_partitions |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_round_robin_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_hpz_partition_size |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_quantized_weights |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_quantized_nontrainable_weights |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_quantized_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zeropp_loco_param |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_log_trace_cache_warnings |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zero_allgather_sequential |
1 | 0 | 0 |
meth |
DeepSpeedEngine.is_sanity_checks_enabled |
1 | 0 | 0 |
meth |
DeepSpeedEngine.dump_state |
1 | 0 | 0 |
meth |
DeepSpeedEngine.gradient_clipping |
1 | 0 | 0 |
meth |
DeepSpeedEngine.dynamic_loss_scale |
1 | 0 | 0 |
meth |
DeepSpeedEngine.initial_dynamic_scale |
1 | 0 | 0 |
meth |
DeepSpeedEngine.dynamic_loss_scale_args |
1 | 0 | 0 |
meth |
DeepSpeedEngine.swap_tensor_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.aio_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.zenflow_config |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_data_types |
1 | 0 | 0 |
meth |
DeepSpeedEngine._optimizer_has_ckpt_event_prologue |
1 | 0 | 0 |
meth |
DeepSpeedEngine._optimizer_has_ckpt_event_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_lr_scheduler |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_checkpointing |
1 | 0 | 0 |
meth |
DeepSpeedEngine._scheduler_from_config |
2 | 0 | 0 |
meth |
DeepSpeedEngine._set_distributed_vars |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_with_arguments |
3 | 0 | 0 |
meth |
DeepSpeedEngine._do_args_sanity_check |
2 | 0 | 0 |
meth |
DeepSpeedEngine._is_supported_optimizer |
2 | 0 | 0 |
meth |
DeepSpeedEngine._supported_optims |
1 | 0 | 0 |
meth |
DeepSpeedEngine._do_sanity_check |
1 | 0 | 0 |
meth |
DeepSpeedEngine._broadcast_model |
1 | 0 | 0 |
meth |
DeepSpeedEngine._set_client_model |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_distributed_model |
2 | 0 | 0 |
meth |
DeepSpeedEngine._check_for_duplicates |
2 | 0 | 0 |
meth |
DeepSpeedEngine._do_optimizer_sanity_check |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_optimizer |
3 | 0 | 0 |
meth |
DeepSpeedEngine._configure_basic_optimizer |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_compression_scheduler |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_random_ltd_scheduler |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_quantization |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_fp16_optimizer |
3 | 0 | 0 |
meth |
DeepSpeedEngine._configure_bf16_optimizer |
2 | 0 | 0 |
meth |
DeepSpeedEngine._configure_zero_optimizer |
2 | 0 | 0 |
meth |
DeepSpeedEngine._return_mics_optimizer |
3 | 0 | 0 |
meth |
DeepSpeedEngine._configure_eigenvalue |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_progressive_layer_drop |
1 | 0 | 0 |
meth |
DeepSpeedEngine._configure_curriculum_scheduler_legacy |
1 | 0 | 0 |
meth |
DeepSpeedEngine.is_map_style_dataset |
2 | 0 | 0 |
meth |
DeepSpeedEngine.is_iterable_style_dataset |
2 | 0 | 0 |
meth |
DeepSpeedEngine.dataloader_drop_last |
1 | 0 | 0 |
meth |
DeepSpeedEngine.deepspeed_io |
8 | 0 | 0 |
meth |
DeepSpeedEngine.train |
2 | 0 | 0 |
meth |
DeepSpeedEngine.eval |
1 | 0 | 0 |
meth |
DeepSpeedEngine._scale_loss_by_gas |
3 | 0 | 0 |
meth |
DeepSpeedEngine._create_module_forward_pre_hook |
1 | 0 | 0 |
meth |
DeepSpeedEngine._create_module_forward_post_hook |
1 | 0 | 0 |
meth |
DeepSpeedEngine._forward_prologue |
3 | 0 | 0 |
meth |
DeepSpeedEngine._forward_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedEngine.forward |
3 | 0 | 0 |
meth |
DeepSpeedEngine._cast_inputs_half |
2 | 0 | 0 |
meth |
DeepSpeedEngine.print_forward_breakdown |
2 | 0 | 0 |
meth |
DeepSpeedEngine.allreduce_gradients |
2 | 0 | 0 |
meth |
DeepSpeedEngine._backward_prologue |
1 | 0 | 0 |
meth |
DeepSpeedEngine._backward_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedEngine._backward_prologue_per_tensor |
2 | 0 | 0 |
meth |
DeepSpeedEngine._backward_post_hook |
1 | 0 | 0 |
meth |
DeepSpeedEngine.no_sync |
1 | 0 | 0 |
meth |
DeepSpeedEngine.scale |
2 | 0 | 0 |
meth |
DeepSpeedEngine.backward |
4 | 0 | 0 |
meth |
DeepSpeedEngine.is_gradient_accumulation_boundary |
1 | 0 | 0 |
meth |
DeepSpeedEngine.set_gradient_accumulation_boundary |
2 | 0 | 0 |
meth |
DeepSpeedEngine.zero_grad |
1 | 0 | 0 |
meth |
DeepSpeedEngine.clip_fp32_gradients |
1 | 0 | 0 |
meth |
DeepSpeedEngine._take_model_step |
3 | 0 | 0 |
meth |
DeepSpeedEngine.step |
2 | 0 | 0 |
meth |
DeepSpeedEngine._start_timers |
2 | 0 | 0 |
meth |
DeepSpeedEngine._stop_timers |
2 | 0 | 0 |
meth |
DeepSpeedEngine._update_wall_clock_timers |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_wall_clock_timers |
1 | 0 | 0 |
meth |
DeepSpeedEngine._autotuning_exit |
1 | 0 | 0 |
meth |
DeepSpeedEngine._write_monitor |
1 | 0 | 0 |
meth |
DeepSpeedEngine._get_optimizer_param |
2 | 0 | 0 |
meth |
DeepSpeedEngine._get_optimizer_loss_scale |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_lr |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_type |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_mom |
1 | 0 | 0 |
meth |
DeepSpeedEngine.get_pld_theta |
1 | 0 | 0 |
meth |
DeepSpeedEngine._report_progress |
2 | 0 | 0 |
meth |
DeepSpeedEngine.allreduce_bucket |
4 | 0 | 0 |
meth |
DeepSpeedEngine.allreduce_and_copy |
4 | 0 | 0 |
meth |
DeepSpeedEngine.allreduce_no_retain |
5 | 0 | 0 |
meth |
DeepSpeedEngine._get_gradients_for_reduction |
1 | 0 | 0 |
meth |
DeepSpeedEngine._reduce_non_expert_gradients |
3 | 0 | 0 |
meth |
DeepSpeedEngine._reduce_expert_gradients |
3 | 0 | 0 |
meth |
DeepSpeedEngine.buffered_allreduce_fallback |
3 | 0 | 0 |
meth |
DeepSpeedEngine.sparse_allreduce_no_retain |
4 | 0 | 0 |
meth |
DeepSpeedEngine.sparse_allreduce_bucket |
4 | 0 | 0 |
meth |
DeepSpeedEngine.sparse_allreduce |
4 | 0 | 0 |
meth |
DeepSpeedEngine.sparse_all_gather |
3 | 0 | 0 |
meth |
DeepSpeedEngine.all_gather_scalar |
3 | 0 | 0 |
meth |
DeepSpeedEngine.module_state_dict |
5 | 0 | 0 |
meth |
DeepSpeedEngine.load_moe_state_dict |
9 | 0 | 0 |
meth |
DeepSpeedEngine.load_module_state_dict |
5 | 0 | 0 |
meth |
DeepSpeedEngine._get_zero_ckpt_prefix |
3 | 0 | 0 |
meth |
DeepSpeedEngine._get_rank_zero_ckpt_name |
6 | 0 | 0 |
meth |
DeepSpeedEngine._get_zero_ckpt_name |
3 | 0 | 0 |
meth |
DeepSpeedEngine._get_ckpt_name |
5 | 0 | 0 |
meth |
DeepSpeedEngine._get_optimizer_ckpt_name |
4 | 0 | 0 |
meth |
DeepSpeedEngine._get_expert_ckpt_name |
6 | 0 | 0 |
meth |
DeepSpeedEngine._get_all_ckpt_names |
3 | 0 | 0 |
meth |
DeepSpeedEngine.load_checkpoint |
8 | 0 | 0 |
meth |
DeepSpeedEngine._load_checkpoint |
8 | 0 | 0 |
meth |
DeepSpeedEngine._load_zero_checkpoint |
4 | 0 | 0 |
meth |
DeepSpeedEngine._get_mp_rank_zero_checkpoint_names |
6 | 0 | 0 |
meth |
DeepSpeedEngine._get_all_zero_checkpoint_names |
4 | 0 | 0 |
meth |
DeepSpeedEngine._get_all_zero_checkpoint_state_dicts |
2 | 0 | 0 |
meth |
DeepSpeedEngine._get_all_zero_checkpoints |
3 | 0 | 0 |
meth |
DeepSpeedEngine._checkpoint_tag_validation |
2 | 0 | 0 |
meth |
DeepSpeedEngine.save_checkpoint |
6 | 0 | 0 |
meth |
DeepSpeedEngine._commit_decoupled_checkpoint |
1 | 0 | 0 |
meth |
DeepSpeedEngine._get_non_moe_state_dict |
2 | 0 | 0 |
meth |
DeepSpeedEngine._save_moe_checkpoint |
5 | 0 | 0 |
meth |
DeepSpeedEngine._create_checkpoint_file |
4 | 0 | 0 |
meth |
DeepSpeedEngine._create_zero_checkpoint_files |
3 | 0 | 0 |
meth |
DeepSpeedEngine._save_checkpoint |
5 | 0 | 0 |
meth |
DeepSpeedEngine._get_buffer_names |
1 | 0 | 0 |
meth |
DeepSpeedEngine._get_param_shape_func |
2 | 0 | 0 |
meth |
DeepSpeedEngine._get_param_fragment_func |
2 | 0 | 0 |
meth |
DeepSpeedEngine._get_zero_frozen_param_attributes |
2 | 0 | 0 |
meth |
DeepSpeedEngine._get_zero_param_shapes |
1 | 0 | 0 |
meth |
DeepSpeedEngine._get_shared_params |
1 | 0 | 0 |
meth |
DeepSpeedEngine._copy_recovery_script |
2 | 0 | 0 |
meth |
DeepSpeedEngine._change_recovery_script_permissions |
2 | 0 | 0 |
meth |
DeepSpeedEngine._save_zero_checkpoint |
3 | 0 | 0 |
meth |
DeepSpeedEngine._replace_module_consolidated_state_dict |
1 | 0 | 0 |
meth |
DeepSpeedEngine._consolidated_16bit_state_dict |
2 | 0 | 0 |
meth |
DeepSpeedEngine._zero3_consolidated_16bit_state_dict |
2 | 0 | 0 |
meth |
DeepSpeedEngine.save_fp16_model |
3 | 0 | 0 |
meth |
DeepSpeedEngine.save_16bit_model |
4 | 0 | 0 |
meth |
DeepSpeedEngine.empty_partition_cache |
1 | 0 | 0 |
meth |
DeepSpeedEngine.compile |
5 | 1 | 0 |
meth |
DeepSpeedEngine.get_compile_time |
1 | 0 | 0 |
prop |
DeepSpeedEngine.communication_data_type |
2 | 0 | 0 |
attr |
DeepSpeedEngine.dont_change_device |
1 | 0 | 0 |
attr |
DeepSpeedEngine.client_optimizer |
1 | 0 | 0 |
attr |
DeepSpeedEngine.client_lr_scheduler |
1 | 0 | 0 |
attr |
DeepSpeedEngine.training_data |
1 | 0 | 0 |
attr |
DeepSpeedEngine.collate_fn |
1 | 0 | 0 |
attr |
DeepSpeedEngine.mpu |
1 | 0 | 0 |
attr |
DeepSpeedEngine.all_to_all_group |
1 | 0 | 0 |
attr |
DeepSpeedEngine.data_parallel_group |
1 | 0 | 0 |
attr |
DeepSpeedEngine.global_steps |
1 | 0 | 0 |
attr |
DeepSpeedEngine.global_samples |
1 | 0 | 0 |
attr |
DeepSpeedEngine.micro_steps |
1 | 0 | 0 |
attr |
DeepSpeedEngine.skipped_steps |
1 | 0 | 0 |
attr |
DeepSpeedEngine.gradient_average |
1 | 0 | 0 |
attr |
DeepSpeedEngine.warn_unscaled_loss |
1 | 0 | 0 |
attr |
DeepSpeedEngine.config |
1 | 0 | 0 |
attr |
DeepSpeedEngine.loaded_checkpoint_mp_world_size |
1 | 0 | 0 |
attr |
DeepSpeedEngine.loaded_checkpoint_dp_world_size |
1 | 0 | 0 |
attr |
DeepSpeedEngine.enable_backward_allreduce |
1 | 0 | 0 |
attr |
DeepSpeedEngine.inside_no_sync_ctxt |
1 | 0 | 0 |
attr |
DeepSpeedEngine.progressive_layer_drop |
1 | 0 | 0 |
attr |
DeepSpeedEngine.eigenvalue |
1 | 0 | 0 |
attr |
DeepSpeedEngine.block_eigenvalue |
1 | 0 | 0 |
attr |
DeepSpeedEngine.gas_boundary_ctr |
1 | 0 | 0 |
attr |
DeepSpeedEngine.dist_backend |
1 | 0 | 0 |
attr |
DeepSpeedEngine.has_moe_layers |
1 | 0 | 0 |
attr |
DeepSpeedEngine.num_experts |
1 | 0 | 0 |
attr |
DeepSpeedEngine.gate_modules |
1 | 0 | 0 |
attr |
DeepSpeedEngine.moe_layers |
1 | 0 | 0 |
attr |
DeepSpeedEngine.use_ds_comm |
1 | 0 | 0 |
attr |
DeepSpeedEngine.checkpoint_engine |
1 | 0 | 0 |
attr |
DeepSpeedEngine.optimizer |
1 | 0 | 0 |
attr |
DeepSpeedEngine.basic_optimizer |
1 | 0 | 0 |
attr |
DeepSpeedEngine.lr_scheduler |
1 | 0 | 0 |
attr |
DeepSpeedEngine.scale_wrt_gas |
1 | 0 | 0 |
attr |
DeepSpeedEngine.losses |
1 | 0 | 0 |
attr |
DeepSpeedEngine.mesh_device |
1 | 0 | 0 |
attr |
DeepSpeedEngine.monitor |
1 | 0 | 0 |
attr |
DeepSpeedEngine.pipeline_parallelism |
1 | 0 | 0 |
attr |
DeepSpeedEngine.module_forward_pre_hook |
1 | 0 | 0 |
attr |
DeepSpeedEngine.module_forward_post_hook |
1 | 0 | 0 |
attr |
DeepSpeedEngine.param_names |
1 | 0 | 0 |
attr |
DeepSpeedEngine.timers |
1 | 0 | 0 |
attr |
DeepSpeedEngine.tput_timer |
1 | 0 | 0 |
attr |
DeepSpeedEngine.torch_autocast_z0_gradscaler |
1 | 0 | 0 |
attr |
DeepSpeedEngine.sparse_tensor_module_names |
1 | 0 | 0 |
attr |
DeepSpeedEngine.save_non_zero_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedEngine.save_zero_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedEngine.engine_timers |
1 | 0 | 0 |
attr |
DeepSpeedEngine.engine_timers_cache |
1 | 0 | 0 |
attr |
DeepSpeedEngine.flatten |
1 | 0 | 0 |
attr |
DeepSpeedEngine.unflatten |
1 | 0 | 0 |
attr |
DeepSpeedEngine.flops_profiler |
1 | 0 | 0 |
attr |
DeepSpeedEngine.training_dataloader |
1 | 0 | 0 |
attr |
DeepSpeedEngine.curriculum_scheduler_legacy |
1 | 0 | 0 |
attr |
DeepSpeedEngine.random_ltd_scheduler |
1 | 0 | 0 |
meth |
EngineTimers.init |
3 | 0 | 0 |
meth |
EngineTimers.active_timers |
1 | 0 | 0 |
attr |
EngineTimers.forward_timers |
1 | 0 | 0 |
attr |
EngineTimers.backward_timers |
1 | 0 | 0 |
attr |
EngineTimers.backward_inner_timers |
1 | 0 | 0 |
attr |
EngineTimers.backward_reduce_timers |
1 | 0 | 0 |
attr |
EngineTimers.step_timers |
1 | 0 | 0 |
attr |
EngineTimers.global_timers |
1 | 0 | 0 |
attr |
EngineTimers.micro_timers |
1 | 0 | 0 |
deepspeed.runtime.fp16.fused_optimizer (82 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FP16_Optimizer.init |
15 | 0 | 0 |
meth |
FP16_Optimizer.initialize_optimizer_states |
1 | 0 | 0 |
meth |
FP16_Optimizer.zero_grad |
2 | 0 | 0 |
meth |
FP16_Optimizer.step_fused_adam |
2 | 0 | 0 |
meth |
FP16_Optimizer.set_lr |
2 | 0 | 0 |
meth |
FP16_Optimizer.get_lr |
1 | 0 | 0 |
meth |
FP16_Optimizer.override_loss_scale |
2 | 0 | 0 |
meth |
FP16_Optimizer._require_avoid_recompute_norm |
3 | 0 | 0 |
meth |
FP16_Optimizer._get_norm_mask_idx |
2 | 0 | 0 |
meth |
FP16_Optimizer.step |
2 | 0 | 0 |
meth |
FP16_Optimizer.unscale_and_clip_grads |
4 | 0 | 0 |
meth |
FP16_Optimizer.backward |
4 | 0 | 0 |
meth |
FP16_Optimizer._update_scale |
2 | 0 | 0 |
meth |
FP16_Optimizer._get_state |
1 | 0 | 0 |
meth |
FP16_Optimizer._set_state |
2 | 0 | 0 |
meth |
FP16_Optimizer._get_param_groups |
1 | 0 | 0 |
meth |
FP16_Optimizer._set_param_groups |
2 | 0 | 0 |
meth |
FP16_Optimizer.state_dict |
1 | 0 | 0 |
meth |
FP16_Optimizer.refresh_fp32_params |
1 | 0 | 0 |
meth |
FP16_Optimizer.load_state_dict |
3 | 0 | 0 |
meth |
FP16_Optimizer.repr |
1 | 0 | 0 |
meth |
FP16_Optimizer._get_loss_scale |
1 | 0 | 0 |
meth |
FP16_Optimizer._set_loss_scale |
2 | 0 | 0 |
attr |
FP16_Optimizer.fused_adam_legacy |
1 | 0 | 0 |
attr |
FP16_Optimizer.timers |
1 | 0 | 0 |
attr |
FP16_Optimizer.has_moe_layers |
1 | 0 | 0 |
attr |
FP16_Optimizer.deepspeed |
1 | 0 | 0 |
attr |
FP16_Optimizer.using_pipeline |
1 | 0 | 0 |
attr |
FP16_Optimizer.low_precision_dtype |
1 | 0 | 0 |
attr |
FP16_Optimizer.loss_scale_config |
1 | 0 | 0 |
attr |
FP16_Optimizer.optimizer |
1 | 0 | 0 |
attr |
FP16_Optimizer.fp16_groups |
1 | 0 | 0 |
attr |
FP16_Optimizer.fp16_groups_flat |
1 | 0 | 0 |
attr |
FP16_Optimizer.fp32_groups_flat |
1 | 0 | 0 |
attr |
FP16_Optimizer.flatten_grad_norm_mask_list |
1 | 0 | 0 |
attr |
FP16_Optimizer.has_executed_step |
1 | 0 | 0 |
attr |
FP16_Optimizer.verbose |
1 | 0 | 0 |
attr |
FP16_Optimizer.custom_loss_scaler |
1 | 0 | 0 |
attr |
FP16_Optimizer.external_loss_scale |
1 | 0 | 0 |
attr |
FP16_Optimizer.clip_grad |
1 | 0 | 0 |
attr |
FP16_Optimizer.norm_type |
1 | 0 | 0 |
attr |
FP16_Optimizer.mpu |
1 | 0 | 0 |
attr |
FP16_Optimizer.overflow |
1 | 0 | 0 |
attr |
FP16_Optimizer.overflow_checker |
1 | 0 | 0 |
attr |
FP16_Optimizer.clip_grad_norm |
1 | 0 | 0 |
attr |
FP16_Optimizer.state |
1 | 0 | 0 |
attr |
FP16_Optimizer.param_groups |
1 | 0 | 0 |
attr |
FP16_Optimizer.loss_scale |
1 | 0 | 0 |
deepspeed.runtime.fp16.loss_scaler (58 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DynamicLossScaler.init |
8 | 0 | 0 |
meth |
DynamicLossScaler.has_overflow_serial |
2 | 0 | 0 |
meth |
DynamicLossScaler._has_inf_or_nan |
1 | 0 | 0 |
meth |
DynamicLossScaler.update_scale |
2 | 0 | 0 |
attr |
DynamicLossScaler.cur_iter |
1 | 0 | 0 |
attr |
DynamicLossScaler.last_overflow_iter |
1 | 0 | 0 |
attr |
DynamicLossScaler.scale_factor |
1 | 0 | 0 |
attr |
DynamicLossScaler.scale_window |
1 | 0 | 0 |
attr |
DynamicLossScaler.min_scale |
1 | 0 | 0 |
attr |
DynamicLossScaler.delayed_shift |
1 | 0 | 0 |
attr |
DynamicLossScaler.cur_hysteresis |
1 | 0 | 0 |
attr |
DynamicLossScaler.consecutive_hysteresis |
1 | 0 | 0 |
attr |
DynamicLossScaler.raise_error_at_min_scale |
1 | 0 | 0 |
attr |
DynamicLossScaler.dynamic |
1 | 0 | 0 |
attr |
DynamicLossScaler.dtype |
1 | 0 | 0 |
func |
to_python_float |
2 | 0 | 0 |
attr |
LOSS_SCALE_PROFILE_DEFAULTS |
1 | 0 | 0 |
func |
CreateLossScaler |
5 | 0 | 0 |
meth |
LossScalerBase.init |
2 | 0 | 0 |
meth |
LossScalerBase.scale_gradient |
4 | 0 | 0 |
meth |
LossScalerBase.update_scale |
2 | 0 | 0 |
meth |
LossScalerBase.scale_loss |
2 | 0 | 0 |
meth |
LossScalerBase.backward |
3 | 0 | 0 |
prop |
LossScalerBase.loss_scale |
1 | 0 | 0 |
attr |
LossScalerBase.cur_scale |
1 | 0 | 0 |
attr |
LossScalerBase.dynamic |
1 | 0 | 0 |
meth |
LossScaler.init |
2 | 0 | 0 |
meth |
LossScaler.has_overflow |
2 | 0 | 0 |
meth |
LossScaler._has_inf_or_nan |
1 | 0 | 0 |
meth |
LossScaleConfig.init |
7 | 2 | 0 |
deepspeed.runtime.fp16.onebit.adam (34 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OnebitAdam.init |
14 | 0 | 0 |
meth |
OnebitAdam.step |
3 | 0 | 0 |
meth |
OnebitAdam.load_state_dict |
2 | 0 | 0 |
attr |
OnebitAdam.eps_mode |
1 | 0 | 0 |
attr |
OnebitAdam.comm_time |
1 | 0 | 0 |
attr |
OnebitAdam.step_time |
1 | 0 | 0 |
attr |
OnebitAdam.ave_step |
1 | 0 | 0 |
attr |
OnebitAdam.bk_time |
1 | 0 | 0 |
attr |
OnebitAdam.deepspeed |
1 | 0 | 0 |
attr |
OnebitAdam.adam_freeze_key |
1 | 0 | 0 |
attr |
OnebitAdam.initialize |
1 | 0 | 0 |
attr |
OnebitAdam.freeze_step |
1 | 0 | 0 |
attr |
OnebitAdam.cuda_aware |
1 | 0 | 0 |
attr |
OnebitAdam.using_pipeline |
1 | 0 | 0 |
attr |
OnebitAdam.comm_backend_name |
1 | 0 | 0 |
attr |
OnebitAdam.comm_backend_handle |
1 | 0 | 0 |
attr |
OnebitAdam.size |
1 | 0 | 0 |
attr |
OnebitAdam.divider |
1 | 0 | 0 |
deepspeed.runtime.fp16.onebit.lamb (48 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OnebitLamb.init |
20 | 0 | 0 |
meth |
OnebitLamb.step |
3 | 0 | 0 |
meth |
OnebitLamb.load_state_dict |
2 | 0 | 0 |
meth |
OnebitLamb.get_lamb_coeffs |
1 | 0 | 0 |
attr |
OnebitLamb.eps_mode |
1 | 0 | 0 |
attr |
OnebitLamb.deepspeed |
1 | 0 | 0 |
attr |
OnebitLamb.lamb_freeze_key |
1 | 0 | 0 |
attr |
OnebitLamb.initialize |
1 | 0 | 0 |
attr |
OnebitLamb.freeze_step |
1 | 0 | 0 |
attr |
OnebitLamb.cuda_aware |
1 | 0 | 0 |
attr |
OnebitLamb.coeff_beta |
1 | 0 | 0 |
attr |
OnebitLamb.factor_max |
1 | 0 | 0 |
attr |
OnebitLamb.factor_min |
1 | 0 | 0 |
attr |
OnebitLamb.factor_threshold |
1 | 0 | 0 |
attr |
OnebitLamb.using_pipeline |
1 | 0 | 0 |
attr |
OnebitLamb.comm_backend_name |
1 | 0 | 0 |
attr |
OnebitLamb.comm_backend_handle |
1 | 0 | 0 |
attr |
OnebitLamb.size |
1 | 0 | 0 |
attr |
OnebitLamb.divider |
1 | 0 | 0 |
attr |
OnebitLamb.exp_avg_flat |
1 | 0 | 0 |
attr |
OnebitLamb.dummy_exp_avg |
1 | 0 | 0 |
attr |
OnebitLamb.corrected_tensor_sizes |
1 | 0 | 0 |
attr |
OnebitLamb.server_chunk_sizes |
1 | 0 | 0 |
attr |
OnebitLamb.worker_errors |
1 | 0 | 0 |
attr |
OnebitLamb.server_errors |
1 | 0 | 0 |
attr |
OnebitLamb.lamb_coeffs |
1 | 0 | 0 |
deepspeed.runtime.fp16.onebit.zoadam (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZeroOneAdam.init |
17 | 0 | 0 |
meth |
ZeroOneAdam.step |
3 | 0 | 0 |
meth |
ZeroOneAdam.load_state_dict |
2 | 0 | 0 |
attr |
ZeroOneAdam.eps_mode |
1 | 0 | 0 |
attr |
ZeroOneAdam.deepspeed |
1 | 0 | 0 |
attr |
ZeroOneAdam.initialize |
1 | 0 | 0 |
attr |
ZeroOneAdam.cuda_aware |
1 | 0 | 0 |
attr |
ZeroOneAdam.using_pipeline |
1 | 0 | 0 |
attr |
ZeroOneAdam.var_freeze_step |
1 | 0 | 0 |
attr |
ZeroOneAdam.var_update_scaler |
1 | 0 | 0 |
attr |
ZeroOneAdam.local_step_scaler |
1 | 0 | 0 |
attr |
ZeroOneAdam.local_step_clipper |
1 | 0 | 0 |
attr |
ZeroOneAdam.freeze_key |
1 | 0 | 0 |
attr |
ZeroOneAdam.reinitial_error_buffer |
1 | 0 | 0 |
attr |
ZeroOneAdam.comm_backend_name |
1 | 0 | 0 |
attr |
ZeroOneAdam.comm_backend_handle |
1 | 0 | 0 |
attr |
ZeroOneAdam.size |
1 | 0 | 0 |
attr |
ZeroOneAdam.divider |
1 | 0 | 0 |
deepspeed.runtime.fp16.unfused_optimizer (66 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FP16_UnfusedOptimizer.init |
12 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.zero_grad |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.step_fused_lamb |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.set_lr |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.get_lr |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.override_loss_scale |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.step |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.unscale_and_clip_grads |
3 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.backward |
4 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._update_scale |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._get_state |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._set_state |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._get_param_groups |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._set_param_groups |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._get_loss_scale |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer._set_loss_scale |
2 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.state_dict |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.refresh_fp32_params |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.load_state_dict |
3 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.repr |
1 | 0 | 0 |
meth |
FP16_UnfusedOptimizer.initialize_optimizer_states |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.fused_lamb_legacy |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.low_precision_dtype |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.loss_scale_config |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.optimizer |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.fp16_groups |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.fp32_groups |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.custom_loss_scaler |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.external_loss_scale |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.verbose |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.clip_grad |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.norm_type |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.mpu |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.overflow |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.overflow_checker |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.clip_grad_norm |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.state |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.param_groups |
1 | 0 | 0 |
attr |
FP16_UnfusedOptimizer.loss_scale |
1 | 0 | 0 |
deepspeed.runtime.hybrid_engine (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedHybridEngine.init |
4 | 0 | 0 |
meth |
DeepSpeedHybridEngine.convert_to_linear_transposed |
2 | 0 | 0 |
meth |
DeepSpeedHybridEngine.new_inference_container |
4 | 0 | 0 |
meth |
DeepSpeedHybridEngine.populate_all_inference_policies |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine._fuse_lora_layer |
2 | 0 | 0 |
meth |
DeepSpeedHybridEngine.fuse_lora_weight |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine._unfuse_lora_layer |
2 | 0 | 0 |
meth |
DeepSpeedHybridEngine.unfuse_lora_weight |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine.unfuse_lora_weight_non_pinned |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine.retake_inference_cache |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine.generate |
3 | 0 | 0 |
meth |
DeepSpeedHybridEngine.create_inference_containers |
3 | 0 | 0 |
meth |
DeepSpeedHybridEngine.create_inference_module |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine._zero3_forward |
2 | 0 | 0 |
meth |
DeepSpeedHybridEngine.eval |
1 | 0 | 0 |
meth |
DeepSpeedHybridEngine.train |
2 | 0 | 0 |
meth |
DeepSpeedHybridEngine.step |
2 | 0 | 0 |
attr |
DeepSpeedHybridEngine.Z3_enabled |
1 | 0 | 0 |
attr |
DeepSpeedHybridEngine.gather_all_layers |
1 | 0 | 0 |
attr |
DeepSpeedHybridEngine.is_lora_fused |
1 | 0 | 0 |
attr |
DeepSpeedHybridEngine.workspace |
1 | 0 | 0 |
deepspeed.runtime.lr_schedules (139 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
override_lr_range_test_params |
3 | 0 | 0 |
func |
get_config_from_args |
2 | 0 | 0 |
meth |
WarmupLR.init |
7 | 6 | 0 |
meth |
WarmupLR.get_lr |
1 | 0 | 0 |
meth |
WarmupLR.get_last_lr |
1 | 0 | 0 |
meth |
WarmupLR.step |
2 | 0 | 0 |
meth |
WarmupLR.state_dict |
1 | 0 | 0 |
meth |
WarmupLR.load_state_dict |
2 | 0 | 0 |
meth |
WarmupLR._get_gamma |
1 | 0 | 0 |
meth |
WarmupLR._format_param |
4 | 0 | 0 |
attr |
WarmupLR.optimizer |
1 | 0 | 0 |
attr |
WarmupLR.min_lrs |
1 | 0 | 0 |
attr |
WarmupLR.max_lrs |
1 | 0 | 0 |
attr |
WarmupLR.delta_lrs |
1 | 0 | 0 |
attr |
WarmupLR.warmup_num_steps |
1 | 0 | 0 |
attr |
WarmupLR.warmup_type |
1 | 0 | 0 |
attr |
WarmupLR.inverse_log_warm_up |
1 | 0 | 0 |
attr |
WarmupLR.last_batch_iteration |
1 | 0 | 0 |
func |
override_warmupLR_params |
3 | 0 | 0 |
meth |
LRRangeTest.init |
7 | 6 | 0 |
meth |
LRRangeTest._staircase_interval |
1 | 0 | 0 |
meth |
LRRangeTest._continuous_interval |
1 | 0 | 0 |
meth |
LRRangeTest._get_increase |
1 | 0 | 0 |
meth |
LRRangeTest.get_lr |
1 | 0 | 0 |
meth |
LRRangeTest.get_last_lr |
1 | 0 | 0 |
meth |
LRRangeTest.step |
2 | 0 | 0 |
meth |
LRRangeTest.state_dict |
1 | 0 | 0 |
meth |
LRRangeTest.load_state_dict |
2 | 0 | 0 |
attr |
LRRangeTest.optimizer |
1 | 0 | 0 |
attr |
LRRangeTest.step_size |
1 | 0 | 0 |
attr |
LRRangeTest.step_rate |
1 | 0 | 0 |
attr |
LRRangeTest.last_batch_iteration |
1 | 0 | 0 |
attr |
LRRangeTest.staircase |
1 | 0 | 0 |
attr |
LRRangeTest.interval_fn |
1 | 0 | 0 |
attr |
LRRangeTest.min_lr |
1 | 0 | 0 |
meth |
WarmupCosineLR.init |
8 | 7 | 0 |
meth |
WarmupCosineLR.get_lr_ratio |
1 | 0 | 0 |
meth |
WarmupCosineLR.step |
2 | 0 | 0 |
meth |
WarmupCosineLR.get_lr |
1 | 0 | 0 |
meth |
WarmupCosineLR.get_last_lr |
1 | 0 | 0 |
meth |
WarmupCosineLR.state_dict |
1 | 0 | 0 |
meth |
WarmupCosineLR.load_state_dict |
2 | 0 | 0 |
meth |
WarmupCosineLR._format_param |
4 | 0 | 0 |
attr |
WarmupCosineLR.optimizer |
1 | 0 | 0 |
attr |
WarmupCosineLR.total_num_steps |
1 | 0 | 0 |
attr |
WarmupCosineLR.last_batch_iteration |
1 | 0 | 0 |
attr |
WarmupCosineLR.cos_min_ratio |
1 | 0 | 0 |
attr |
WarmupCosineLR.warmup_type |
1 | 0 | 0 |
attr |
WarmupCosineLR.warmup_min_ratio |
1 | 0 | 0 |
attr |
WarmupCosineLR.warmup_num_steps |
1 | 0 | 0 |
attr |
WarmupCosineLR.inverse_log_warm_up |
1 | 0 | 0 |
attr |
WarmupCosineLR.org_lrs |
1 | 0 | 0 |
func |
get_lr_from_config |
2 | 0 | 0 |
meth |
OneCycle.init |
15 | 0 | 0 |
meth |
OneCycle._initialize_cycle |
6 | 0 | 0 |
meth |
OneCycle._initialize_lr |
6 | 0 | 0 |
meth |
OneCycle._initialize_momentum |
6 | 0 | 0 |
meth |
OneCycle._get_scale_factor |
1 | 0 | 0 |
meth |
OneCycle._get_cycle_mom |
1 | 0 | 0 |
meth |
OneCycle._get_cycle_lr |
1 | 0 | 0 |
meth |
OneCycle._get_decay_mom |
2 | 0 | 0 |
meth |
OneCycle._get_decay_lr |
2 | 0 | 0 |
meth |
OneCycle.get_lr |
1 | 0 | 0 |
meth |
OneCycle.get_mom |
1 | 0 | 0 |
meth |
OneCycle.get_last_lr |
1 | 0 | 0 |
meth |
OneCycle.step |
2 | 0 | 0 |
meth |
OneCycle.state_dict |
1 | 0 | 0 |
meth |
OneCycle.load_state_dict |
2 | 0 | 0 |
attr |
OneCycle.optimizer |
1 | 0 | 0 |
attr |
OneCycle.cycle_momentum |
1 | 0 | 0 |
attr |
OneCycle.last_batch_iteration |
1 | 0 | 0 |
func |
override_1cycle_params |
3 | 0 | 0 |
func |
get_torch_optimizer |
2 | 0 | 0 |
func |
override_params |
3 | 0 | 0 |
func |
update_lr |
3 | 0 | 0 |
func |
add_tuning_arguments |
2 | 0 | 0 |
meth |
WarmupDecayLR.init |
8 | 7 | 0 |
meth |
WarmupDecayLR._get_gamma |
1 | 0 | 0 |
attr |
WarmupDecayLR.total_num_steps |
1 | 0 | 0 |
func |
parse_arguments |
1 | 0 | 0 |
deepspeed.runtime.model_checkpointing.config (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_checkpoint_config |
2 | 0 | 0 |
func |
get_checkpoint_writer_config |
2 | 0 | 0 |
deepspeed.runtime.model_checkpointing.data_parallel_writer_factory (23 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DataParallelWriterFactory.init |
3 | 0 | 0 |
meth |
DataParallelWriterFactory.create_config |
3 | 0 | 0 |
meth |
DataParallelWriterFactory._create_config |
3 | 0 | 0 |
meth |
DataParallelWriterFactory._get_expert_data_parallel_config |
1 | 0 | 0 |
meth |
DataParallelWriterFactory._get_expert_parallel_write_for_2d |
1 | 0 | 0 |
meth |
DataParallelWriterFactory._get_data_parallel_config |
1 | 0 | 0 |
meth |
DataParallelWriterFactory._get_parallel_write_for_3d |
1 | 0 | 0 |
meth |
DataParallelWriterFactory._get_slice_writers |
3 | 0 | 0 |
meth |
DataParallelWriterFactory._assign_resources_to_tensor_slice |
4 | 0 | 0 |
meth |
DataParallelWriterFactory._get_parallel_write_for_ddp |
3 | 0 | 0 |
deepspeed.runtime.model_checkpointing.utils (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
create_universal_parallel_info |
3 | 0 | 0 |
func |
create_data_parallel_writer_config |
5 | 0 | 0 |
deepspeed.runtime.model_checkpointing.writer_factory (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
CheckpointWriterFactory.init |
4 | 0 | 0 |
meth |
CheckpointWriterFactory.create_writer |
3 | 0 | 0 |
meth |
CheckpointWriterFactory.release_writer |
1 | 0 | 0 |
meth |
CheckpointWriterFactory._setup_for_aio |
2 | 0 | 0 |
meth |
CheckpointWriterFactory._setup_for_gds |
2 | 0 | 0 |
deepspeed.runtime.pipe.engine (148 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PipelineEngine.init |
4 | 0 | 0 |
meth |
PipelineEngine.set_has_attention_mask |
2 | 0 | 0 |
meth |
PipelineEngine._build_data_iter |
2 | 0 | 0 |
meth |
PipelineEngine._exec_reduce_tied_grads |
1 | 0 | 0 |
meth |
PipelineEngine._exec_reduce_grads |
1 | 0 | 0 |
meth |
PipelineEngine._bf16_reduce_grads |
1 | 0 | 0 |
meth |
PipelineEngine._reserve_pipe_buffers |
2 | 0 | 0 |
meth |
PipelineEngine.reset_activation_shape |
1 | 0 | 0 |
meth |
PipelineEngine.train_batch |
2 | 0 | 0 |
meth |
PipelineEngine.eval_batch |
7 | 0 | 0 |
meth |
PipelineEngine.set_train_batch_size |
2 | 0 | 0 |
meth |
PipelineEngine.is_first_stage |
1 | 0 | 0 |
meth |
PipelineEngine.is_last_stage |
1 | 0 | 0 |
meth |
PipelineEngine.get_pipeline_parallel_rank |
1 | 0 | 0 |
meth |
PipelineEngine._reduce_outputs |
5 | 0 | 0 |
meth |
PipelineEngine._bcast_pipe_scalar |
4 | 0 | 0 |
meth |
PipelineEngine._aggregate_total_loss |
1 | 0 | 0 |
meth |
PipelineEngine.set_dataloader |
2 | 0 | 0 |
meth |
PipelineEngine.set_dataiterator |
2 | 0 | 0 |
meth |
PipelineEngine.set_batch_fn |
2 | 0 | 0 |
meth |
PipelineEngine.is_gradient_accumulation_boundary |
1 | 0 | 0 |
meth |
PipelineEngine.log_for_device |
2 | 0 | 0 |
meth |
PipelineEngine.tput_log |
2 | 0 | 0 |
meth |
PipelineEngine._next_batch |
1 | 0 | 0 |
meth |
PipelineEngine._exec_forward_pass |
2 | 0 | 0 |
meth |
PipelineEngine._exec_backward_pass |
2 | 0 | 0 |
meth |
PipelineEngine._exec_load_micro_batch |
2 | 0 | 0 |
meth |
PipelineEngine._send_tensor_meta |
3 | 0 | 0 |
meth |
PipelineEngine._recv_tensor_meta |
2 | 0 | 0 |
meth |
PipelineEngine._exec_send_activations |
2 | 0 | 0 |
meth |
PipelineEngine._exec_send_grads |
2 | 0 | 0 |
meth |
PipelineEngine._exec_recv_activations |
2 | 0 | 0 |
meth |
PipelineEngine._exec_recv_grads |
2 | 0 | 0 |
meth |
PipelineEngine._exec_optimizer_step |
2 | 0 | 0 |
meth |
PipelineEngine._allocate_zeros |
3 | 0 | 0 |
meth |
PipelineEngine._allocate_buffer |
4 | 0 | 0 |
meth |
PipelineEngine._allocate_or_extend_buffers |
4 | 0 | 0 |
meth |
PipelineEngine.forward |
3 | 0 | 0 |
meth |
PipelineEngine.backward |
3 | 0 | 0 |
meth |
PipelineEngine.step |
3 | 0 | 0 |
meth |
PipelineEngine.module_state_dict |
2 | 0 | 0 |
meth |
PipelineEngine.load_module_state_dict |
5 | 0 | 0 |
meth |
PipelineEngine._exec_schedule |
2 | 0 | 0 |
meth |
PipelineEngine.get_additional_losses |
1 | 0 | 0 |
attr |
PipelineEngine.DTYPE_TO_ID |
1 | 0 | 0 |
attr |
PipelineEngine.enable_backward_allreduce |
1 | 0 | 0 |
attr |
PipelineEngine.has_bool_tensors |
1 | 0 | 0 |
attr |
PipelineEngine.eval_return_logits |
1 | 0 | 0 |
attr |
PipelineEngine.outputs |
1 | 0 | 0 |
attr |
PipelineEngine.using_bf16_optimizer |
1 | 0 | 0 |
attr |
PipelineEngine.pipeline_enable_backward_allreduce |
1 | 0 | 0 |
attr |
PipelineEngine.log_batch_step_id |
1 | 0 | 0 |
attr |
PipelineEngine.micro_batch_size |
1 | 0 | 0 |
attr |
PipelineEngine.micro_batches |
1 | 0 | 0 |
attr |
PipelineEngine.grid |
1 | 0 | 0 |
attr |
PipelineEngine.global_rank |
1 | 0 | 0 |
attr |
PipelineEngine.num_stages |
1 | 0 | 0 |
attr |
PipelineEngine.stage_id |
1 | 0 | 0 |
attr |
PipelineEngine.prev_stage |
1 | 0 | 0 |
attr |
PipelineEngine.next_stage |
1 | 0 | 0 |
attr |
PipelineEngine.data_iterator |
1 | 0 | 0 |
attr |
PipelineEngine.batch_fn |
1 | 0 | 0 |
attr |
PipelineEngine.batch_timer |
1 | 0 | 0 |
attr |
PipelineEngine.is_pipe_parallel |
1 | 0 | 0 |
attr |
PipelineEngine.is_data_parallel |
1 | 0 | 0 |
attr |
PipelineEngine.is_model_parallel |
1 | 0 | 0 |
attr |
PipelineEngine.is_pipe_partitioned |
1 | 0 | 0 |
attr |
PipelineEngine.is_grad_partitioned |
1 | 0 | 0 |
attr |
PipelineEngine.num_pipe_buffers |
1 | 0 | 0 |
attr |
PipelineEngine.pipe_buffers |
1 | 0 | 0 |
attr |
PipelineEngine.pipe_recv_buf |
1 | 0 | 0 |
attr |
PipelineEngine.grad_layer |
1 | 0 | 0 |
attr |
PipelineEngine.meta_buffer |
1 | 0 | 0 |
attr |
PipelineEngine.first_output_send |
1 | 0 | 0 |
attr |
PipelineEngine.first_gradient_send |
1 | 0 | 0 |
attr |
PipelineEngine.pipe_partition_input_meta_cache |
1 | 0 | 0 |
attr |
PipelineEngine.pipe_partition_output_meta_cache |
1 | 0 | 0 |
attr |
PipelineEngine.pipe_partition_grad_meta_cache |
1 | 0 | 0 |
attr |
PipelineEngine.grad_partition_grad_layer_meta_cache |
1 | 0 | 0 |
attr |
PipelineEngine.loss |
1 | 0 | 0 |
attr |
PipelineEngine.total_loss |
1 | 0 | 0 |
attr |
PipelineEngine.total_additional_losses |
1 | 0 | 0 |
attr |
PipelineEngine.agg_loss |
1 | 0 | 0 |
attr |
PipelineEngine.dp_group_loss |
1 | 0 | 0 |
attr |
PipelineEngine.agg_train_loss |
1 | 0 | 0 |
attr |
PipelineEngine.agg_additional_losses |
1 | 0 | 0 |
attr |
PipelineEngine.has_attention_mask |
1 | 0 | 0 |
attr |
PipelineEngine.dynamic_shape |
1 | 0 | 0 |
attr |
PipelineEngine.loss_model |
1 | 0 | 0 |
func |
is_even |
2 | 0 | 0 |
deepspeed.runtime.pipe.module (105 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
LayerSpec.init |
4 | 0 | 0 |
meth |
LayerSpec.repr |
1 | 0 | 0 |
meth |
LayerSpec.build |
2 | 0 | 0 |
attr |
LayerSpec.typename |
1 | 0 | 0 |
attr |
LayerSpec.module_args |
1 | 0 | 0 |
attr |
LayerSpec.module_kwargs |
1 | 0 | 0 |
attr |
LayerSpec.global_rank |
1 | 0 | 0 |
meth |
TiedLayerSpec.init |
7 | 0 | 0 |
attr |
TiedLayerSpec.key |
1 | 0 | 0 |
attr |
TiedLayerSpec.forward_fn |
1 | 0 | 0 |
attr |
TiedLayerSpec.tied_weight_attr |
1 | 0 | 0 |
meth |
PipelineModule.init |
13 | 0 | 0 |
meth |
PipelineModule._precompute_checkpointable_values |
1 | 0 | 0 |
meth |
PipelineModule._build |
1 | 0 | 0 |
meth |
PipelineModule._get_frozen_parameter_names |
2 | 0 | 0 |
meth |
PipelineModule._count_layer_params |
1 | 0 | 0 |
meth |
PipelineModule._find_layer_type |
2 | 0 | 0 |
meth |
PipelineModule.forward |
2 | 0 | 0 |
meth |
PipelineModule._partition_layers |
2 | 0 | 0 |
meth |
PipelineModule.allreduce_tied_weight_gradients |
1 | 0 | 0 |
meth |
PipelineModule.get_tied_weights_and_groups |
1 | 0 | 0 |
meth |
PipelineModule._synchronize_tied_weights |
1 | 0 | 0 |
meth |
PipelineModule._index_tied_modules |
1 | 0 | 0 |
meth |
PipelineModule.partitions |
1 | 0 | 0 |
meth |
PipelineModule.stage_owner |
2 | 0 | 0 |
meth |
PipelineModule._set_bounds |
3 | 0 | 0 |
meth |
PipelineModule.set_checkpoint_interval |
2 | 0 | 0 |
meth |
PipelineModule.topology |
1 | 0 | 0 |
meth |
PipelineModule.mpu |
1 | 0 | 0 |
meth |
PipelineModule.num_pipeline_stages |
1 | 0 | 0 |
meth |
PipelineModule.ckpt_prefix |
3 | 0 | 0 |
meth |
PipelineModule.ckpt_layer_path |
3 | 0 | 0 |
meth |
PipelineModule.ckpt_layer_path_list |
3 | 0 | 0 |
meth |
PipelineModule.save_state_dict |
4 | 0 | 0 |
meth |
PipelineModule.load_state_dir |
4 | 0 | 0 |
meth |
PipelineModule._is_checkpointable |
2 | 0 | 0 |
meth |
PipelineModule.get_additional_losses |
1 | 0 | 0 |
meth |
PipelineModule.compile |
3 | 0 | 0 |
attr |
PipelineModule.micro_offset |
1 | 0 | 0 |
attr |
PipelineModule.loss_fn |
1 | 0 | 0 |
attr |
PipelineModule.checkpointable_layers |
1 | 0 | 0 |
attr |
PipelineModule.seed_layers |
1 | 0 | 0 |
attr |
PipelineModule.seed_fn |
1 | 0 | 0 |
attr |
PipelineModule.base_seed |
1 | 0 | 0 |
attr |
PipelineModule.world_group |
1 | 0 | 0 |
attr |
PipelineModule.global_rank |
1 | 0 | 0 |
attr |
PipelineModule.world_size |
1 | 0 | 0 |
attr |
PipelineModule.local_rank |
1 | 0 | 0 |
attr |
PipelineModule.stage_id |
1 | 0 | 0 |
attr |
PipelineModule.forward_funcs |
1 | 0 | 0 |
attr |
PipelineModule.fwd_map |
1 | 0 | 0 |
attr |
PipelineModule.tied_modules |
1 | 0 | 0 |
attr |
PipelineModule.tied_weight_attrs |
1 | 0 | 0 |
attr |
PipelineModule.activation_checkpoint_interval |
1 | 0 | 0 |
attr |
PipelineModule.activation_checkpoint_func |
1 | 0 | 0 |
attr |
PipelineModule.is_checkpointable_results |
1 | 0 | 0 |
attr |
PipelineModule.is_checkpointable_results_interval |
1 | 0 | 0 |
attr |
PipelineModule.tied_comms |
1 | 0 | 0 |
attr |
PipelineModule.dynamic_shape |
1 | 0 | 0 |
attr |
PipelineModule.num_stages |
1 | 0 | 0 |
deepspeed.runtime.pipe.p2p (12 missing, 2 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
init_process_groups |
2 | 0 | 0 |
func |
recv |
4 | 0 | 0 |
func |
send |
4 | 0 | 0 |
func |
send_obj |
3 | 2 | 1 |
func |
recv_obj |
2 | 2 | 1 |
func |
wait |
1 | 0 | 0 |
deepspeed.runtime.pipe.schedule (48 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TrainSchedule.steps |
1 | 0 | 0 |
meth |
TrainSchedule.num_pipe_buffers |
1 | 0 | 0 |
meth |
TrainSchedule._step_to_micro_batch |
2 | 0 | 0 |
meth |
TrainSchedule._even_step_forward_id |
2 | 0 | 0 |
meth |
TrainSchedule._odd_step_forward_id |
2 | 0 | 0 |
meth |
TrainSchedule._even_step_backward_id |
2 | 0 | 0 |
meth |
TrainSchedule._odd_step_backward_id |
2 | 0 | 0 |
meth |
DataParallelSchedule.steps |
1 | 0 | 0 |
meth |
DataParallelSchedule.num_pipe_buffers |
1 | 0 | 0 |
meth |
PipeInstruction.init |
2 | 0 | 0 |
meth |
PipeInstruction.repr |
1 | 0 | 0 |
attr |
PipeInstruction.name |
1 | 0 | 0 |
attr |
PipeInstruction.kwargs |
1 | 0 | 0 |
meth |
BufferOpInstruction.init |
3 | 0 | 0 |
meth |
PipeSchedule.init |
4 | 0 | 0 |
meth |
PipeSchedule.steps |
1 | 0 | 0 |
meth |
PipeSchedule.num_pipe_buffers |
1 | 0 | 0 |
meth |
PipeSchedule._valid_micro_batch |
2 | 0 | 0 |
meth |
PipeSchedule._valid_stage |
2 | 0 | 0 |
meth |
PipeSchedule._buffer_idx |
2 | 0 | 0 |
meth |
PipeSchedule.iter |
1 | 0 | 0 |
meth |
PipeSchedule.next |
1 | 0 | 0 |
prop |
PipeSchedule.stage |
1 | 0 | 0 |
prop |
PipeSchedule.num_stages |
1 | 0 | 0 |
prop |
PipeSchedule.num_micro_batches |
1 | 0 | 0 |
prop |
PipeSchedule.is_first_stage |
1 | 0 | 0 |
prop |
PipeSchedule.is_last_stage |
1 | 0 | 0 |
attr |
PipeSchedule.micro_batches |
1 | 0 | 0 |
attr |
PipeSchedule.stages |
1 | 0 | 0 |
attr |
PipeSchedule.stage_id |
1 | 0 | 0 |
attr |
PipeSchedule.prev_stage |
1 | 0 | 0 |
attr |
PipeSchedule.next_stage |
1 | 0 | 0 |
meth |
InferenceSchedule.steps |
1 | 0 | 0 |
meth |
InferenceSchedule.num_pipe_buffers |
1 | 0 | 0 |
deepspeed.runtime.pipe.topology (88 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PipeDataParallelTopology.init |
3 | 0 | 0 |
meth |
PipelineParallelGrid.init |
3 | 0 | 0 |
meth |
PipelineParallelGrid.get_stage_id |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_data_parallel_id |
1 | 0 | 0 |
meth |
PipelineParallelGrid._build_p2p_groups |
1 | 0 | 0 |
meth |
PipelineParallelGrid._is_grid_valid |
1 | 0 | 0 |
meth |
PipelineParallelGrid.stage_to_global |
3 | 0 | 0 |
meth |
PipelineParallelGrid.topology |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_global_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_pipe_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_pipeline_model_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_pipe_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_pipeline_model_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_pipe_parallel_group |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_data_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_data_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_data_parallel_group |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_data_parallel_group_ranks |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_model_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_model_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_model_parallel_group |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_slice_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_tensor_model_parallel_rank |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_slice_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_tensor_model_parallel_world_size |
1 | 0 | 0 |
meth |
PipelineParallelGrid.get_slice_parallel_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.global_rank |
1 | 0 | 0 |
attr |
PipelineParallelGrid.world_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.data_parallel_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.pipe_parallel_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.model_parallel_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.slice_parallel_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.stage_id |
1 | 0 | 0 |
attr |
PipelineParallelGrid.data_parallel_id |
1 | 0 | 0 |
attr |
PipelineParallelGrid.ds_model_proc_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.ds_model_rank |
1 | 0 | 0 |
attr |
PipelineParallelGrid.dp_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.dp_groups |
1 | 0 | 0 |
attr |
PipelineParallelGrid.is_first_stage |
1 | 0 | 0 |
attr |
PipelineParallelGrid.is_last_stage |
1 | 0 | 0 |
attr |
PipelineParallelGrid.p2p_groups |
1 | 0 | 0 |
attr |
PipelineParallelGrid.pp_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.pp_proc_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.pipe_groups |
1 | 0 | 0 |
attr |
PipelineParallelGrid.mp_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.model_groups |
1 | 0 | 0 |
attr |
PipelineParallelGrid.ds_model_world_size |
1 | 0 | 0 |
attr |
PipelineParallelGrid.dp_proc_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.slice_group |
1 | 0 | 0 |
attr |
PipelineParallelGrid.slice_proc_group |
1 | 0 | 0 |
meth |
ProcessTopology.init |
3 | 0 | 0 |
meth |
ProcessTopology.get_rank |
2 | 0 | 0 |
meth |
ProcessTopology.get_axis_names |
1 | 0 | 0 |
meth |
ProcessTopology.get_rank_repr |
5 | 0 | 0 |
meth |
ProcessTopology.get_dim |
2 | 0 | 0 |
meth |
ProcessTopology.get_coord |
2 | 0 | 0 |
meth |
ProcessTopology.get_axis_comm_lists |
2 | 0 | 0 |
meth |
ProcessTopology.filter_match |
2 | 0 | 0 |
meth |
ProcessTopology.get_axis_list |
3 | 0 | 0 |
meth |
ProcessTopology.world_size |
1 | 0 | 0 |
meth |
ProcessTopology.str |
1 | 0 | 0 |
attr |
ProcessTopology.axes |
1 | 0 | 0 |
attr |
ProcessTopology.dims |
1 | 0 | 0 |
attr |
ProcessTopology.ProcessCoord |
1 | 0 | 0 |
attr |
ProcessTopology.mapping |
1 | 0 | 0 |
meth |
PipeModelDataParallelTopology.init |
4 | 0 | 0 |
deepspeed.runtime.precision_config (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_float16_config |
2 | 0 | 0 |
meth |
DeepSpeedFP16Config._validate_loss_scale |
2 | 0 | 0 |
meth |
DeepSpeedFP16Config.initial_dynamic_scale |
1 | 0 | 0 |
meth |
DeepSpeedFP16Config.dynamic_loss_scale_args |
1 | 0 | 0 |
func |
get_bfloat16_config |
2 | 0 | 0 |
deepspeed.runtime.progressive_layer_drop (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ProgressiveLayerDrop.init |
3 | 0 | 0 |
meth |
ProgressiveLayerDrop.get_state |
1 | 0 | 0 |
meth |
ProgressiveLayerDrop.get_theta |
1 | 0 | 0 |
meth |
ProgressiveLayerDrop.update_state |
2 | 0 | 0 |
attr |
ProgressiveLayerDrop.theta |
1 | 0 | 0 |
attr |
ProgressiveLayerDrop.gamma |
1 | 0 | 0 |
attr |
ProgressiveLayerDrop.current_theta |
1 | 0 | 0 |
deepspeed.runtime.quantize (44 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
Quantizer.init |
10 | 0 | 0 |
meth |
Quantizer.any_precision_switch |
1 | 0 | 0 |
meth |
Quantizer.quantize |
5 | 0 | 0 |
meth |
Quantizer.step |
1 | 0 | 0 |
meth |
Quantizer.quantize_highbit |
3 | 0 | 0 |
meth |
Quantizer.quantize_tenary |
2 | 0 | 0 |
meth |
Quantizer.quantize_binary |
2 | 0 | 0 |
meth |
Quantizer.mixed_fp16_quantize |
4 | 0 | 0 |
meth |
Quantizer.compute_quantization |
4 | 0 | 0 |
meth |
Quantizer.update_fp16_ratio |
1 | 0 | 0 |
attr |
Quantizer.q_groups |
1 | 0 | 0 |
attr |
Quantizer.q_mixed_fp16 |
1 | 0 | 0 |
attr |
Quantizer.q_change_ratio |
1 | 0 | 0 |
attr |
Quantizer.q_type |
1 | 0 | 0 |
attr |
Quantizer.qsteps |
1 | 0 | 0 |
attr |
Quantizer.quantize_real_ratio |
1 | 0 | 0 |
attr |
Quantizer.q_verbose |
1 | 0 | 0 |
attr |
Quantizer.q_eigenvalue |
1 | 0 | 0 |
attr |
Quantizer.use_quantizer_kernel |
1 | 0 | 0 |
attr |
Quantizer.q_rounding |
1 | 0 | 0 |
attr |
Quantizer.layer_num |
1 | 0 | 0 |
deepspeed.runtime.sequence_parallel.parallel_state_sp (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_sequence_data_parallel_group |
1 | 0 | 0 |
func |
get_sequence_data_parallel_rank |
1 | 0 | 0 |
func |
get_sequence_parallel_rank |
1 | 0 | 0 |
func |
get_sequence_data_parallel_world_size |
1 | 0 | 0 |
func |
get_sequence_parallel_group |
1 | 0 | 0 |
func |
get_sequence_parallel_world_size |
1 | 0 | 0 |
deepspeed.runtime.sequence_parallel.ulysses_sp (126 missing, 2 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TiledFusedLogitsLoss.forward |
10 | 1 | 0 |
meth |
TiledFusedLogitsLoss.backward |
3 | 1 | 0 |
meth |
UlyssesSPAttentionHF.init |
12 | 11 | 0 |
meth |
UlyssesSPAttentionHF._combine_local_sequences |
4 | 1 | 0 |
meth |
UlyssesSPAttentionHF._partition_global_sequence |
2 | 1 | 0 |
meth |
UlyssesSPAttentionHF.forward |
8 | 8 | 2 |
meth |
UlyssesSPAttentionHF.register_with_transformers |
9 | 0 | 0 |
attr |
UlyssesSPAttentionHF.attn |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.process_group |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.world_size |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.sp_rank |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.batch_size |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.seq_length_is_variable |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.local_seq_length |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.global_seq_length |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.disable_in_eval |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.attn_head_size |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.attn_head_count |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.global_kv_head_count |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.num_hidden_layers |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.skip_all_but_last_attention_debug_mode |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.rotating_layer_counter |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.local_q_head_count |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.kv_replication_factor |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.required_query_shape |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.required_key_value_shape |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.required_context_shape |
1 | 0 | 0 |
attr |
UlyssesSPAttentionHF.local_kv_head_count |
1 | 0 | 0 |
meth |
TiledMLP.forward |
7 | 1 | 0 |
meth |
TiledMLP.backward |
3 | 1 | 0 |
meth |
UlyssesSPDataLoaderAdapter.init |
6 | 2 | 0 |
meth |
UlyssesSPDataLoaderAdapter.len |
1 | 0 | 0 |
meth |
UlyssesSPDataLoaderAdapter.iter |
1 | 0 | 0 |
meth |
UlyssesSPDataLoaderAdapter.next |
1 | 0 | 0 |
meth |
UlyssesSPDataLoaderAdapter.refill |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.dl |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.sp_rank |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.sp_group |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.sp_world_size |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.device |
1 | 0 | 0 |
attr |
UlyssesSPDataLoaderAdapter.iter |
1 | 0 | 0 |
meth |
AutogradComputeMLP.forward |
5 | 1 | 0 |
meth |
AutogradComputeMLP.backward |
3 | 1 | 0 |
meth |
TiledLoss.forward |
7 | 1 | 0 |
meth |
TiledLoss.backward |
3 | 1 | 0 |
meth |
UlyssesSPFwdLossBwdWithLogits.init |
6 | 0 | 0 |
meth |
UlyssesSPFwdLossBwdWithLogits.sp_fwd_loss_bwd |
2 | 1 | 0 |
meth |
UlyssesSPFwdLossBwdWithLogits.forward |
2 | 0 | 0 |
meth |
UlyssesSPFwdLossBwdWithLogits.compute_loss |
3 | 0 | 0 |
meth |
UlyssesSPFwdLossBwdWithLogits.backward |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.model |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.model_unwrapped |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.device |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.num_loss_logit_shards |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.kwargs |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.sp_group |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.sp_world_size |
1 | 0 | 0 |
attr |
UlyssesSPFwdLossBwdWithLogits.sp_rank |
1 | 0 | 0 |
meth |
SequenceTiledCompute.forward |
12 | 1 | 0 |
meth |
SequenceTiledCompute.backward |
3 | 1 | 0 |
func |
sequence_tiled_compute |
10 | 0 | 0 |
deepspeed.runtime.sparse_tensor (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SparseTensor.init |
2 | 0 | 0 |
meth |
SparseTensor.to_coo_tensor |
1 | 0 | 0 |
meth |
SparseTensor.type |
1 | 0 | 0 |
meth |
SparseTensor.to_dense |
1 | 0 | 0 |
meth |
SparseTensor.sparse_size |
1 | 0 | 0 |
meth |
SparseTensor.add |
2 | 0 | 0 |
meth |
SparseTensor.str |
1 | 0 | 0 |
meth |
SparseTensor.repr |
1 | 0 | 0 |
attr |
SparseTensor.orig_dense_tensor |
1 | 0 | 0 |
attr |
SparseTensor.is_sparse |
1 | 0 | 0 |
attr |
SparseTensor.dtype |
1 | 0 | 0 |
attr |
SparseTensor.dense_size |
1 | 0 | 0 |
attr |
SparseTensor.indices |
1 | 0 | 0 |
attr |
SparseTensor.values |
1 | 0 | 0 |
deepspeed.runtime.state_dict_factory (85 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MegatronSDLoader.init |
4 | 0 | 0 |
meth |
MegatronSDLoader.merge_query_key_value |
3 | 0 | 0 |
meth |
MegatronSDLoader.split_query_key_value |
5 | 0 | 0 |
meth |
MegatronSDLoader.merge_state_dict |
7 | 0 | 0 |
meth |
MegatronSDLoader.split_state_dict |
7 | 0 | 0 |
meth |
MegatronSDLoader.sanity_check |
2 | 0 | 0 |
meth |
MegatronSDLoader.get_checkpoint_version |
2 | 0 | 0 |
meth |
SDLoaderBase.init |
4 | 0 | 0 |
meth |
SDLoaderBase.load |
9 | 0 | 0 |
meth |
SDLoaderBase.get_merge_state_dicts |
3 | 0 | 0 |
meth |
SDLoaderBase.get_split_state_dict |
3 | 0 | 0 |
meth |
SDLoaderBase._choose_module_key |
2 | 0 | 0 |
meth |
SDLoaderBase.get_module |
2 | 0 | 0 |
meth |
SDLoaderBase.set_module |
3 | 0 | 0 |
meth |
SDLoaderBase.check_ckpt_list |
1 | 0 | 0 |
meth |
SDLoaderBase.merge_state_dict |
7 | 0 | 0 |
meth |
SDLoaderBase.split_state_dict |
7 | 0 | 0 |
meth |
SDLoaderBase.sanity_check |
2 | 0 | 0 |
attr |
SDLoaderBase.module_key |
1 | 0 | 0 |
attr |
SDLoaderBase.ckpt_list |
1 | 0 | 0 |
attr |
SDLoaderBase.version |
1 | 0 | 0 |
attr |
SDLoaderBase.checkpoint_engine |
1 | 0 | 0 |
meth |
SDLoaderFactory.get_sd_loader_json |
3 | 0 | 0 |
meth |
SDLoaderFactory.get_sd_loader |
5 | 0 | 0 |
deepspeed.runtime.superoffload.superoffload_stage3 (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SuperOffloadOptimizer_Stage3.init |
7 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._create_fp16_sub_groups |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._optimizer_step |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3.reduce_independent_p_g_buckets_and_remove_grads |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._reassign_or_swap_out_partitioned_parameters |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._reassign_or_swap_out_partitioned_parameters_async |
3 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3.step |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._wait_for_async_operations |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._wait_for_single_async_result |
3 | 1 | 0 |
meth |
SuperOffloadOptimizer_Stage3._sync_cpu_optimizer_step |
7 | 4 | 0 |
meth |
SuperOffloadOptimizer_Stage3._handle_overflow_rollback |
1 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3._handle_gradient_clipping |
2 | 0 | 0 |
meth |
SuperOffloadOptimizer_Stage3.check_clip_grads |
2 | 0 | 0 |
attr |
SuperOffloadOptimizer_Stage3.sub_group_to_param_num |
1 | 0 | 0 |
attr |
SuperOffloadOptimizer_Stage3.params_in_ipg_bucket_buffer |
1 | 0 | 0 |
attr |
SuperOffloadOptimizer_Stage3.async_cpuadam_num |
1 | 0 | 0 |
attr |
SuperOffloadOptimizer_Stage3.max_grad_numel |
1 | 0 | 0 |
attr |
SuperOffloadOptimizer_Stage3.superoffload_cpu_optimizer |
1 | 0 | 0 |
deepspeed.runtime.superoffload.superoffload_utils (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
SuperOffloadCPUOptimizer.max_grad_numel |
1 | 0 | 0 |
attr |
SuperOffloadCPUOptimizer.mp_context |
1 | 0 | 0 |
attr |
SuperOffloadCPUOptimizer.param_queue |
1 | 0 | 0 |
attr |
SuperOffloadCPUOptimizer.result_queue |
1 | 0 | 0 |
attr |
SuperOffloadCPUOptimizer.cpuadam_process |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.aio_config (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_aio_config |
2 | 0 | 0 |
deepspeed.runtime.swap_tensor.async_swapper (49 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
AsyncTensorSwapper.init |
4 | 0 | 0 |
meth |
AsyncTensorSwapper.has_buffers |
1 | 0 | 0 |
meth |
AsyncTensorSwapper.add_buffers |
2 | 0 | 0 |
meth |
AsyncTensorSwapper.get_timer_names |
1 | 0 | 0 |
meth |
AsyncTensorSwapper.release_buffers |
1 | 0 | 0 |
meth |
AsyncTensorSwapper.swap_out_tensors |
3 | 0 | 0 |
meth |
AsyncTensorSwapper._report_statistics |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._swap_out_tensor |
3 | 0 | 0 |
meth |
AsyncTensorSwapper._make_swap_space |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._io_aligned_numel |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._allocate_buffer |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._flush_ready_buffers |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._flush_buffers_until_complete |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._swap_out_ready_buffers |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._wait_for_swap_complete |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._get_buffer |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._get_current_buffer |
1 | 0 | 0 |
meth |
AsyncTensorSwapper._start_timer |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._stop_timer |
2 | 0 | 0 |
meth |
AsyncTensorSwapper._log_timers |
3 | 0 | 0 |
attr |
AsyncTensorSwapper.free_buffer_index |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.swapping_buffer_index |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.ready_buffer_index |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.current_buffer_index |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.all_buffers |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.aio_handle |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.numel_alignment |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.max_numel |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.num_pending_swaps |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.timers |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.timer_names |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.num_elements_swapped |
1 | 0 | 0 |
attr |
AsyncTensorSwapper.dtype |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.optimizer_utils (154 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OptimizerStateSwapInfo.init |
4 | 0 | 0 |
meth |
OptimizerStateSwapInfo.numel |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.has_gradients |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo._add_tensors |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.add_state_tensors |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.num_tensors |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.device |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.dtype |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.release_memory |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_compute_tensors |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_swap_paths |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_swap_buffers_and_paths |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_or_create_gradient_paths |
3 | 0 | 0 |
meth |
OptimizerStateSwapInfo.set_swap_buffers |
3 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_swap_gradient_buffers |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_swap_gradient_paths |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.get_unpinned_state_tensors |
1 | 0 | 0 |
meth |
OptimizerStateSwapInfo.read_unswapped_gradients |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.write_unswapped_gradients |
2 | 0 | 0 |
meth |
OptimizerStateSwapInfo.release_unswapped_gradients |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.tensors |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.param_id |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.swap_folder |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.swapped_gradients |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.unswapped_gradients |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.tensor_numel |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.tensor_dtype |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.tensor_device |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.has_state_tensors |
1 | 0 | 0 |
attr |
OptimizerStateSwapInfo.swap_buffers |
1 | 0 | 0 |
meth |
SwapTensorContext.init |
3 | 0 | 0 |
meth |
SwapTensorContext.release_memory |
1 | 0 | 0 |
meth |
SwapTensorContext.set_buffers |
3 | 0 | 0 |
attr |
SwapTensorContext.compute_tensor |
1 | 0 | 0 |
attr |
SwapTensorContext.swap_tensor |
1 | 0 | 0 |
attr |
SwapTensorContext.swap_path |
1 | 0 | 0 |
meth |
FlattenedTensorSwapInfo.init |
4 | 0 | 0 |
attr |
FlattenedTensorSwapInfo.path |
1 | 0 | 0 |
attr |
FlattenedTensorSwapInfo.offset |
1 | 0 | 0 |
attr |
FlattenedTensorSwapInfo.length |
1 | 0 | 0 |
meth |
OptimizerSwapper.parameter_id |
2 | 0 | 0 |
meth |
OptimizerSwapper.init |
9 | 0 | 0 |
meth |
OptimizerSwapper.purge_state |
1 | 0 | 0 |
meth |
OptimizerSwapper.is_swappable_tensor |
3 | 0 | 0 |
meth |
OptimizerSwapper.init_timers |
1 | 0 | 0 |
meth |
OptimizerSwapper.log_timers |
1 | 0 | 0 |
meth |
OptimizerSwapper.pre_backward |
1 | 0 | 0 |
meth |
OptimizerSwapper.post_backward |
1 | 0 | 0 |
meth |
OptimizerSwapper._flush_gradient_swapper |
2 | 0 | 0 |
meth |
OptimizerSwapper._swap_out_gradients |
5 | 0 | 0 |
meth |
OptimizerSwapper._initialize_from_swapped_fp16_params |
6 | 0 | 0 |
meth |
OptimizerSwapper._swap_in_fp16_params |
5 | 0 | 0 |
meth |
OptimizerSwapper._swap_out_fp16_params |
5 | 0 | 0 |
meth |
OptimizerSwapper._initialize_parameters |
4 | 0 | 0 |
meth |
OptimizerSwapper._get_swap_paths |
3 | 0 | 0 |
meth |
OptimizerSwapper._swap_out_unpinned_tensors |
5 | 0 | 0 |
meth |
OptimizerSwapper._adjust_for_misaligned_lengths |
3 | 0 | 0 |
meth |
OptimizerSwapper._retrieve_unswapped_grad_partitions |
3 | 0 | 0 |
meth |
OptimizerSwapper._get_state_tensors |
2 | 0 | 0 |
meth |
OptimizerSwapper._update_param_state_info |
3 | 0 | 0 |
meth |
OptimizerSwapper._create_param_swap_info |
3 | 0 | 0 |
meth |
OptimizerSwapper._get_param_swap_info |
2 | 0 | 0 |
meth |
OptimizerSwapper._start_timer |
2 | 0 | 0 |
meth |
OptimizerSwapper._stop_timer |
2 | 0 | 0 |
meth |
OptimizerSwapper._log_timers |
3 | 0 | 0 |
meth |
OptimizerSwapper._io_aligned_numel |
2 | 0 | 0 |
attr |
OptimizerSwapper.swap_config |
1 | 0 | 0 |
attr |
OptimizerSwapper.aio_config |
1 | 0 | 0 |
attr |
OptimizerSwapper.swap_params_info |
1 | 0 | 0 |
attr |
OptimizerSwapper.swap_element_size |
1 | 0 | 0 |
attr |
OptimizerSwapper.swap_folder |
1 | 0 | 0 |
attr |
OptimizerSwapper.optimizer |
1 | 0 | 0 |
attr |
OptimizerSwapper.min_aio_bytes |
1 | 0 | 0 |
attr |
OptimizerSwapper.aligned_bytes |
1 | 0 | 0 |
attr |
OptimizerSwapper.numel_alignment |
1 | 0 | 0 |
attr |
OptimizerSwapper.largest_numel |
1 | 0 | 0 |
attr |
OptimizerSwapper.dtype |
1 | 0 | 0 |
attr |
OptimizerSwapper.swap_buffer_manager |
1 | 0 | 0 |
attr |
OptimizerSwapper.timers |
1 | 0 | 0 |
attr |
OptimizerSwapper.timer_names |
1 | 0 | 0 |
attr |
OptimizerSwapper.print_exclude_list |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.partitioned_optimizer_swapper (50 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
PartitionedOptimizerSwapper.init |
9 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.initialize_parameters |
3 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.initialize_from_swapped_fp16_params |
5 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.flush_gradients |
1 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.release_swap_buffers |
2 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.swap_in_optimizer_state |
3 | 0 | 0 |
meth |
PartitionedOptimizerSwapper._swap_out_optimizer_state |
2 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.writeback_optimizer_state_and_gradients |
4 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.swap_out_optimizer_state |
3 | 0 | 0 |
meth |
PartitionedOptimizerSwapper.swap_out_gradients |
4 | 0 | 0 |
meth |
PartitionedOptimizerSwapper._swap_in_parameter |
4 | 0 | 0 |
meth |
PartitionedOptimizerSwapper._swap_in_pinned_gradients |
4 | 0 | 0 |
meth |
PartitionedOptimizerSwapper._swap_in_gradients |
4 | 0 | 0 |
attr |
PartitionedOptimizerSwapper.aio_handle |
1 | 0 | 0 |
attr |
PartitionedOptimizerSwapper.gradient_swapper |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.partitioned_param_swapper (76 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
print_rank_0 |
4 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.init |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.available_swap_in_buffers |
1 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._configure_aio |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.swappable_tensor |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.get_path |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._get_swap_paths |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._get_swap_buffers |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._track_numel |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._allocate_and_return_buffers_for_swap_in |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.synchronize_writes |
1 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.synchronize_reads |
1 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.remove_partition_and_release_buffers |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._swap_out |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.swap_out_and_release |
4 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._update_inflight_swap_in |
4 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.swap_in |
4 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.swap_into_buffer |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.get_buffer |
3 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.reserve_available_buffers |
1 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.release_reserved_buffers |
1 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._io_aligned_numel |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper._is_io_aligned |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.reserve_partitioned_swap_space |
2 | 0 | 0 |
meth |
AsyncPartitionedParameterSwapper.swap_out_partitioned_params |
3 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.dtype |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.id_to_path |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.param_id_to_buffer_id |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.param_id_to_swap_buffer |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.param_id_to_numel |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.pending_writes |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.pending_reads |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.inflight_params |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.inflight_swap_in_buffers |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.inflight_numel |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.available_params |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.available_numel |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.partitioned_swap_buffer |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.partitioned_swap_pool |
1 | 0 | 0 |
attr |
AsyncPartitionedParameterSwapper.invalid_buffer |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.pipelined_optimizer_swapper (60 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OptimizerSwapOp.init |
7 | 0 | 0 |
meth |
OptimizerSwapOp.is_parameter |
2 | 0 | 0 |
meth |
OptimizerSwapOp.wait |
1 | 0 | 0 |
attr |
OptimizerSwapOp.aio_handle |
1 | 0 | 0 |
attr |
OptimizerSwapOp.read_op |
1 | 0 | 0 |
attr |
OptimizerSwapOp.param_info |
1 | 0 | 0 |
attr |
OptimizerSwapOp.allocated_buffers |
1 | 0 | 0 |
attr |
OptimizerSwapOp.state_buffers |
1 | 0 | 0 |
attr |
OptimizerSwapOp.wait_required |
1 | 0 | 0 |
attr |
OptimizerSwapOp.num_ops |
1 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.init |
9 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.initialize_parameters |
3 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.initialize_from_swapped_fp16_params |
5 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.flush_gradients |
1 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.swap_in_optimizer_state |
3 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.swap_out_optimizer_state |
3 | 0 | 0 |
meth |
PipelinedOptimizerSwapper.swap_out_gradients |
4 | 0 | 0 |
meth |
PipelinedOptimizerSwapper._complete_swap_out |
2 | 0 | 0 |
meth |
PipelinedOptimizerSwapper._swap_out_optimizer_state |
4 | 0 | 0 |
meth |
PipelinedOptimizerSwapper._swap_in_optimizer_state |
3 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.write_aio_handle |
1 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.read_aio_handle |
1 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.gradient_swapper |
1 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.async_swap_in |
1 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.async_swap_out |
1 | 0 | 0 |
attr |
PipelinedOptimizerSwapper.swap_ops |
1 | 0 | 0 |
deepspeed.runtime.swap_tensor.utils (88 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SwapBuffer.init |
2 | 0 | 0 |
meth |
SwapBuffer.reset |
1 | 0 | 0 |
meth |
SwapBuffer.insert_tensor |
4 | 0 | 0 |
meth |
SwapBuffer.allocate_tensor |
4 | 0 | 0 |
meth |
SwapBuffer.has_space |
2 | 0 | 0 |
meth |
SwapBuffer.get_swap_tensors |
1 | 0 | 0 |
meth |
SwapBuffer.get_swap_paths |
1 | 0 | 0 |
meth |
SwapBuffer.get_compute_tensors |
1 | 0 | 0 |
meth |
SwapBuffer.get_num_elem |
1 | 0 | 0 |
meth |
SwapBuffer.get_swap_tensor |
2 | 0 | 0 |
meth |
SwapBuffer.get_compute_tensor |
2 | 0 | 0 |
meth |
SwapBuffer.get_swap_path |
2 | 0 | 0 |
attr |
SwapBuffer.buffer |
1 | 0 | 0 |
meth |
SwapBufferManager.init |
4 | 0 | 0 |
meth |
SwapBufferManager.allocate |
4 | 0 | 0 |
meth |
SwapBufferManager.allocate_all |
3 | 0 | 0 |
meth |
SwapBufferManager.free |
2 | 0 | 0 |
attr |
SwapBufferManager.num_elems |
1 | 0 | 0 |
attr |
SwapBufferManager.count |
1 | 0 | 0 |
attr |
SwapBufferManager.dtype |
1 | 0 | 0 |
attr |
SwapBufferManager.all_buffers |
1 | 0 | 0 |
attr |
SwapBufferManager.free_buffer_index |
1 | 0 | 0 |
attr |
SwapBufferManager.used_buffer_index |
1 | 0 | 0 |
attr |
SwapBufferManager.gigabytes |
1 | 0 | 0 |
meth |
SwapBufferPool.init |
2 | 0 | 0 |
meth |
SwapBufferPool.reset |
1 | 0 | 0 |
meth |
SwapBufferPool.allocate_tensor |
4 | 0 | 0 |
meth |
SwapBufferPool.insert_tensor |
4 | 0 | 0 |
meth |
SwapBufferPool.get_swap_tensors |
1 | 0 | 0 |
meth |
SwapBufferPool.get_swap_paths |
1 | 0 | 0 |
meth |
SwapBufferPool.get_compute_tensors |
1 | 0 | 0 |
meth |
SwapBufferPool.has_space |
2 | 0 | 0 |
meth |
SwapBufferPool.swap_out |
3 | 0 | 0 |
meth |
SwapBufferPool.swap_in |
3 | 0 | 0 |
meth |
SwapBufferPool._get_current_buffer |
1 | 0 | 0 |
meth |
SwapBufferPool._get_used_buffers |
1 | 0 | 0 |
attr |
SwapBufferPool.buffers |
1 | 0 | 0 |
attr |
SwapBufferPool.current_index |
1 | 0 | 0 |
func |
print_object |
4 | 0 | 0 |
func |
get_sized_buffers |
3 | 0 | 0 |
func |
get_sized_buffer |
3 | 0 | 0 |
func |
swap_in_tensors |
4 | 0 | 0 |
func |
swap_out_tensors |
4 | 0 | 0 |
deepspeed.runtime.tensor_parallel.config (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TPTrainingConfig.get_partition_config_object |
1 | 0 | 0 |
func |
get_tensor_parallel_config |
2 | 0 | 0 |
deepspeed.runtime.tensor_parallel.init_utils (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
record_tp_model_init_args |
5 | 0 | 0 |
func |
merge_tp_model_init_into_config |
5 | 1 | 0 |
func |
tp_group_world_size |
3 | 0 | 0 |
deepspeed.runtime.tensor_parallel.tp_manager (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TpTrainingManager.init |
4 | 0 | 0 |
meth |
TpTrainingManager._initialize_config |
2 | 0 | 0 |
meth |
TpTrainingManager._apply_policies |
2 | 0 | 0 |
meth |
TpTrainingManager._apply_injection_policy |
3 | 0 | 0 |
meth |
TpTrainingManager._initialize_tp_config |
2 | 0 | 0 |
meth |
TpTrainingManager._get_model_config_generate |
1 | 0 | 0 |
attr |
TpTrainingManager.module |
1 | 0 | 0 |
attr |
TpTrainingManager.config |
1 | 0 | 0 |
deepspeed.runtime.torch_autocast (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
autocast_if_enabled |
2 | 0 | 0 |
func |
init_autocast_params |
4 | 3 | 0 |
deepspeed.runtime.utils (203 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_global_norm_of_tensors |
6 | 0 | 0 |
func |
partition_balanced |
3 | 0 | 0 |
func |
filter_empty_parameters |
2 | 0 | 0 |
meth |
noop_context.init |
1 | 0 | 0 |
meth |
noop_context.enter |
1 | 0 | 0 |
meth |
noop_context.exit |
4 | 0 | 0 |
func |
all_gather_dp_groups |
6 | 0 | 0 |
func |
partition_uniform |
3 | 0 | 0 |
meth |
PartitionedTensor.init |
4 | 0 | 0 |
meth |
PartitionedTensor.from_meta |
5 | 0 | 0 |
meth |
PartitionedTensor.partition_tensor |
2 | 0 | 0 |
meth |
PartitionedTensor.full |
2 | 0 | 0 |
meth |
PartitionedTensor.to_meta |
1 | 0 | 0 |
meth |
PartitionedTensor.data |
1 | 0 | 0 |
meth |
PartitionedTensor.local_size |
1 | 0 | 0 |
meth |
PartitionedTensor.full_size |
1 | 0 | 0 |
attr |
PartitionedTensor.group |
1 | 0 | 0 |
attr |
PartitionedTensor.num_parts |
1 | 0 | 0 |
attr |
PartitionedTensor.rank |
1 | 0 | 0 |
attr |
PartitionedTensor.orig_size |
1 | 0 | 0 |
attr |
PartitionedTensor.orig_device |
1 | 0 | 0 |
attr |
PartitionedTensor.even_split |
1 | 0 | 0 |
func |
register_output_backward_hooks |
4 | 0 | 0 |
func |
clip_grad_norm |
5 | 0 | 0 |
func |
get_weight_norm |
4 | 0 | 0 |
func |
see_memory_usage |
3 | 0 | 0 |
func |
copy_to_device |
4 | 0 | 0 |
func |
get_global_norm |
2 | 0 | 0 |
func |
get_ma_status |
1 | 0 | 0 |
func |
offload_adam_states |
5 | 2 | 0 |
meth |
DummyOptim.init |
2 | 0 | 0 |
attr |
DummyOptim.param_groups |
1 | 0 | 0 |
func |
get_flattened_grad_norm |
5 | 0 | 0 |
func |
get_checkpoint_folder_size |
4 | 0 | 0 |
func |
memory_status |
4 | 0 | 0 |
func |
is_model_parallel_parameter |
2 | 1 | 0 |
func |
graph_process |
5 | 0 | 0 |
func |
get_only_unique_item |
2 | 0 | 0 |
func |
get_norm_with_moe_layers_fast |
3 | 0 | 0 |
meth |
CheckOverflow.init |
5 | 0 | 0 |
meth |
CheckOverflow.check_using_norm |
3 | 0 | 0 |
meth |
CheckOverflow.check |
2 | 0 | 0 |
meth |
CheckOverflow.has_overflow_serial |
2 | 0 | 0 |
meth |
CheckOverflow.has_overflow |
3 | 0 | 0 |
meth |
CheckOverflow._has_inf_or_nan |
3 | 0 | 0 |
attr |
CheckOverflow.mpu |
1 | 0 | 0 |
attr |
CheckOverflow.params |
1 | 0 | 0 |
attr |
CheckOverflow.zero_reduce_scatter |
1 | 0 | 0 |
attr |
CheckOverflow.deepspeed |
1 | 0 | 0 |
attr |
CheckOverflow.has_moe_params |
1 | 0 | 0 |
func |
empty_cache |
1 | 0 | 0 |
func |
get_inactive_params |
2 | 0 | 0 |
func |
get_norm_with_moe_layers |
5 | 0 | 0 |
func |
reload_adam_states |
4 | 1 | 0 |
func |
move_to_device |
4 | 0 | 0 |
func |
all_gather_into_tensor_dp_groups |
4 | 0 | 0 |
func |
set_random_seed |
2 | 0 | 0 |
func |
clip_tensors_by_global_norm |
7 | 0 | 0 |
func |
get_grad_zeros |
3 | 0 | 0 |
func |
maybe_loss_for_backward |
2 | 1 | 0 |
func |
mask_nan_or_inf_with_val_inplace |
4 | 0 | 0 |
meth |
OutputBackwardHookManager.init |
3 | 0 | 0 |
meth |
OutputBackwardHookManager._make_backward_hook |
2 | 0 | 0 |
meth |
OutputBackwardHookManager._traverse_and_register_hooks |
3 | 0 | 0 |
meth |
OutputBackwardHookManager.register_hooks_on_outputs |
2 | 0 | 0 |
meth |
OutputBackwardHookManager.remove_hooks |
1 | 0 | 0 |
meth |
OutputBackwardHookManager.reset |
1 | 0 | 0 |
attr |
OutputBackwardHookManager.preprocess_once_fn |
1 | 0 | 0 |
attr |
OutputBackwardHookManager.preprocess_per_tensor_fn |
1 | 0 | 0 |
attr |
OutputBackwardHookManager.preprocess_done |
1 | 0 | 0 |
attr |
OutputBackwardHookManager.hook_handles |
1 | 0 | 0 |
func |
noop_decorator |
2 | 0 | 0 |
func |
get_tensor_bytes |
2 | 0 | 0 |
func |
ensure_directory_exists |
2 | 0 | 0 |
func |
prefix_sum_inc |
2 | 0 | 0 |
func |
call_to_str |
4 | 0 | 0 |
func |
align_dense_tensors |
3 | 0 | 0 |
meth |
TLinear.init |
3 | 0 | 0 |
meth |
TLinear._fwd |
2 | 0 | 0 |
meth |
TLinear._fwd_bias_add |
2 | 0 | 0 |
meth |
TLinear.forward |
2 | 0 | 0 |
attr |
TLinear.name |
1 | 0 | 0 |
attr |
TLinear.bias |
1 | 0 | 0 |
deepspeed.runtime.weight_quantizer (39 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
WeightQuantization.init |
3 | 0 | 0 |
meth |
WeightQuantization.quantize_data |
5 | 0 | 0 |
meth |
WeightQuantization.is_mlp |
3 | 0 | 0 |
meth |
WeightQuantization.is_qkv |
2 | 0 | 0 |
meth |
WeightQuantization.Quantize |
6 | 0 | 0 |
meth |
WeightQuantization.merge_layer_scales |
2 | 0 | 0 |
meth |
WeightQuantization.merge_scales |
1 | 0 | 0 |
meth |
WeightQuantization.merge_scales_split |
2 | 0 | 0 |
meth |
WeightQuantization.sd_quantize_megatron |
4 | 0 | 0 |
meth |
WeightQuantization.model_quantize |
5 | 0 | 0 |
attr |
WeightQuantization.dense_scales |
1 | 0 | 0 |
attr |
WeightQuantization.qkv_scales |
1 | 0 | 0 |
attr |
WeightQuantization.mlp4hh_scales |
1 | 0 | 0 |
attr |
WeightQuantization.mlph4h_scales |
1 | 0 | 0 |
attr |
WeightQuantization.mlp_extra_grouping |
1 | 0 | 0 |
attr |
WeightQuantization.mp_size |
1 | 0 | 0 |
deepspeed.runtime.zenflow.engine (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
sync_zenflow_optimizer_lr |
2 | 1 | 0 |
func |
is_zenflow_update_boundary |
2 | 1 | 0 |
func |
zenflow_step |
3 | 1 | 0 |
deepspeed.runtime.zenflow.engine_stage3 (29 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
update_selected_channels |
4 | 1 | 0 |
func |
partition_grads |
4 | 3 | 0 |
func |
zenflow_cpu_optimizer_overlap_step |
4 | 0 | 0 |
func |
zenflow_backward_epilogue |
2 | 1 | 0 |
func |
step |
3 | 0 | 0 |
func |
zenflow_cpu_optimizer_step |
2 | 1 | 0 |
func |
wait_last_update_and_copy |
3 | 0 | 0 |
func |
initialize_optimizer_states |
2 | 1 | 0 |
func |
log_selective_optimizer_timers |
2 | 1 | 0 |
func |
configure_zenflow |
3 | 0 | 0 |
func |
zenflow_backward_prologue |
2 | 1 | 0 |
func |
unscale_and_clip_grads |
5 | 0 | 0 |
func |
sync_fp32_param_from_gpu |
2 | 1 | 0 |
func |
selective_optimizer_step |
2 | 1 | 0 |
deepspeed.runtime.zenflow.zenflow_config (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZenFlowConfig.validate_fields |
1 | 0 | 0 |
deepspeed.runtime.zenflow.zenflow_stage_1_and_2 (59 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
ZenFlowZeroOptimizerSequential.init |
3 | 0 | 0 |
meth |
ZenFlowZeroOptimizerSequential.zenflow_cpu_optimizer_step |
2 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.init |
6 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.create |
2 | 0 | 0 |
meth |
ZenFlowZeroOptimizer._configure_zenflow |
2 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.is_zenflow_select_boundary |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.sync_fp32_param_from_gpu |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.update_selected_channels |
4 | 0 | 0 |
meth |
ZenFlowZeroOptimizer._process_selected_fp32_groups_grad |
4 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.average_tensor |
3 | 2 | 0 |
meth |
ZenFlowZeroOptimizer.backward |
3 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.log_selective_optimizer_timers |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizer._sync_selective_optimizer_lr |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizer._selective_optimizer_step |
2 | 0 | 0 |
meth |
ZenFlowZeroOptimizer.selective_optimizer_step |
2 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.micro_step |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.full_warm_up_rounds |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.offload_selective_optimizer |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.pt_reserved_cores_perc |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.start_optimizer_process |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.zf_stage3 |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.selective_optimizer |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizer.num_total_param |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.init |
3 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.initialize_optimizer_states |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel._get_offload_gradient_dict |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.get_overlap_step_state |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.async_inplace_copy_grad_to_fp32_buffer_from_gpu |
2 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.wait_last_update_and_copy |
1 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.zenflow_cpu_optimizer_step |
3 | 0 | 0 |
meth |
ZenFlowZeroOptimizerParallel.step |
2 | 0 | 0 |
attr |
ZenFlowZeroOptimizerParallel.process_optimizer_established |
1 | 0 | 0 |
attr |
ZenFlowZeroOptimizerParallel.first_update_round_after_warmup |
1 | 0 | 0 |
deepspeed.runtime.zenflow.zenflow_utils (11 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
disable_accelerator |
1 | 0 | 0 |
func |
start_optimizer_process |
2 | 0 | 0 |
func |
all_tensors_equal |
2 | 0 | 0 |
func |
zenflow_optimizer_process |
6 | 0 | 0 |
deepspeed.runtime.zero.config (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
read_zero_config_deprecated |
2 | 0 | 0 |
meth |
DeepSpeedZeroConfig.overlap_comm_valid |
1 | 0 | 0 |
meth |
DeepSpeedZeroConfig.offload_ratio_check |
1 | 0 | 0 |
func |
get_zero_config |
2 | 0 | 0 |
deepspeed.runtime.zero.contiguous_memory_allocator (53 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
print_rank_0 |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator.init |
4 | 0 | 0 |
meth |
ContiguousMemoryAllocator.allocate_tensor |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator.assign_to_param |
5 | 0 | 0 |
meth |
ContiguousMemoryAllocator.release_tensor |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator.release_tensor_with_id |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator.print_allocation |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator.max_allocated |
1 | 0 | 0 |
meth |
ContiguousMemoryAllocator._reset_param_data |
1 | 0 | 0 |
meth |
ContiguousMemoryAllocator._unassign_params |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator._release_tensor |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator._consolidate_address |
3 | 0 | 0 |
meth |
ContiguousMemoryAllocator._defragment_memory |
1 | 0 | 0 |
meth |
ContiguousMemoryAllocator._replace_old_address_with_new |
3 | 0 | 0 |
meth |
ContiguousMemoryAllocator._get_new_tensor_address |
2 | 0 | 0 |
meth |
ContiguousMemoryAllocator._get_new_tensor |
3 | 0 | 0 |
meth |
ContiguousMemoryAllocator._largest_contiguous |
1 | 0 | 0 |
meth |
ContiguousMemoryAllocator._mark_as_occupied |
3 | 0 | 0 |
attr |
ContiguousMemoryAllocator.buffer |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.contiguous_sizes |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.tensor_addresses |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.tensor_sizes |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.tensor_ids |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.tensor_map |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.id_to_params |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.total_size |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.total_free |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.largest_contiguous |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.max_allocated |
1 | 0 | 0 |
attr |
ContiguousMemoryAllocator.count |
1 | 0 | 0 |
deepspeed.runtime.zero.leaf_module_config (3 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedZeroLeafModuleConfig._coerce_container_types |
2 | 0 | 0 |
meth |
DeepSpeedZeroLeafModuleConfig._validate_entries |
1 | 0 | 0 |
deepspeed.runtime.zero.linear (19 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
LinearModuleForZeroStage3.bias |
1 | 0 | 0 |
meth |
LinearFunctionForZeroStage3.forward |
5 | 0 | 0 |
meth |
LinearFunctionForZeroStage3.backward |
3 | 0 | 0 |
attr |
autocast_custom_fwd |
1 | 0 | 0 |
attr |
autocast_custom_bwd |
1 | 0 | 0 |
func |
print_rank_0 |
4 | 0 | 0 |
func |
zero3_linear_wrap |
4 | 0 | 0 |
deepspeed.runtime.zero.mics (57 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MiCS_AllGatherCoalescedHandle.init |
5 | 4 | 0 |
meth |
MiCS_AllGatherCoalescedHandle.wait |
2 | 1 | 0 |
func |
has_hierarchical_all_gather_groups |
2 | 1 | 0 |
meth |
MiCS_Offload._convert_to_zero_parameters |
4 | 0 | 0 |
meth |
MiCS_Optimizer.init |
8 | 0 | 0 |
meth |
MiCS_Optimizer.initialize_ds_offload |
3 | 0 | 0 |
meth |
MiCS_Optimizer.allreduce_mics_shard_grads |
3 | 1 | 0 |
meth |
MiCS_Optimizer.load_state_dict |
6 | 0 | 0 |
attr |
MiCS_Optimizer.dp_process_group |
1 | 0 | 0 |
attr |
MiCS_Optimizer.partition_count |
1 | 0 | 0 |
meth |
MiCS_Init.init |
12 | 0 | 0 |
meth |
MiCS_Init._convert_to_deepspeed_param |
2 | 0 | 0 |
meth |
MiCS_Init._pre_all_gather |
3 | 0 | 0 |
meth |
MiCS_Init._flat_all_gather_with_coalescing_manager |
3 | 0 | 0 |
meth |
MiCS_Init._hierarchical_all_gather_params |
3 | 0 | 0 |
meth |
MiCS_Init.get_partition_dp_group |
2 | 0 | 0 |
meth |
MiCS_Init.get_partition_rank |
1 | 0 | 0 |
prop |
MiCS_Init.num_partitions |
1 | 0 | 0 |
attr |
MiCS_Init.mics_comm_groups |
1 | 0 | 0 |
attr |
MiCS_Init.ds_process_group |
1 | 0 | 0 |
deepspeed.runtime.zero.mics_utils (6 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
create_mics_comm_groups |
5 | 0 | 0 |
func |
scale_tensors |
3 | 2 | 0 |
deepspeed.runtime.zero.muon.muon_optimizer (5 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
MuonWithAuxAdam.init |
3 | 0 | 0 |
meth |
MuonWithAuxAdam.step |
2 | 0 | 0 |
deepspeed.runtime.zero.muon.original_muon (37 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
muon_update |
6 | 0 | 0 |
meth |
Muon.init |
5 | 0 | 0 |
meth |
Muon.step |
2 | 0 | 0 |
meth |
MuonWithAuxAdam.init |
2 | 0 | 0 |
meth |
MuonWithAuxAdam.step |
2 | 0 | 0 |
func |
zeropower_via_newtonschulz5 |
3 | 1 | 0 |
func |
adam_update |
7 | 0 | 0 |
meth |
SingleDeviceMuon.init |
5 | 0 | 0 |
meth |
SingleDeviceMuon.step |
2 | 0 | 0 |
meth |
SingleDeviceMuonWithAuxAdam.init |
2 | 0 | 0 |
meth |
SingleDeviceMuonWithAuxAdam.step |
2 | 0 | 0 |
deepspeed.runtime.zero.offload_config (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
DeepSpeedZeroOffloadOptimizerConfig.set_pipeline |
1 | 0 | 0 |
deepspeed.runtime.zero.offload_states (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
offload_optimizer_states |
5 | 0 | 0 |
func |
reload_optimizer_states |
4 | 0 | 0 |
func |
reload_adam_states |
4 | 1 | 0 |
func |
get_state_devices |
3 | 2 | 0 |
func |
offload_adam_states |
5 | 2 | 0 |
deepspeed.runtime.zero.parameter_offload (81 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
FWD_MODULE_STACK |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.init |
19 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.partition_all_parameters |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.get_param_coordinator |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.empty_partition_cache |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._convert_to_zero_parameters |
4 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.destroy |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._remove_module_hooks |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.setup_zero_stage3_hooks |
1 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.mark_persistent_parameters |
3 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._register_deepspeed_module |
3 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.pre_sub_module_forward_function |
2 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.post_sub_module_forward_function |
2 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.pre_sub_module_backward_function |
2 | 0 | 0 |
meth |
DeepSpeedZeRoOffload.post_sub_module_backward_function |
2 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._set_z3_leaf_modules_by_threshold |
3 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._get_granularity_recursively |
2 | 0 | 0 |
meth |
DeepSpeedZeRoOffload._set_leaf_by_threshold_preorder |
3 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.module |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.timers |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.zenflow |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.dp_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.offload_device |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.offload_param_pin_memory |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.zero_param_parallel_group |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.zero_quantized_weights |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.zero_quantized_nontrainable_weights |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.log_trace_cache_warnings |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.param_numel_persistence_threshold |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.model_persistence_threshold |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.persistent_parameters |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.fast_sharding_for_leaf_module |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.param_coordinator |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.forward_hooks |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.backward_hooks |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.min_granularity_value |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.min_granularity_layer |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.granularity_info |
1 | 0 | 0 |
attr |
DeepSpeedZeRoOffload.z3_leaf_layers |
1 | 0 | 0 |
meth |
ZeROOrderedDict.init |
4 | 0 | 0 |
meth |
ZeROOrderedDict.reduce |
1 | 0 | 0 |
meth |
ZeROOrderedDict.getitem |
2 | 0 | 0 |
deepspeed.runtime.zero.partition_parameters (202 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_allgather_dtype |
3 | 0 | 0 |
meth |
Init.init |
17 | 0 | 0 |
meth |
Init._update_persist_config |
2 | 0 | 0 |
meth |
Init._zero_init_param |
2 | 0 | 0 |
meth |
Init._convert_to_zero_parameters |
2 | 0 | 0 |
meth |
Init._validate_remote_device |
3 | 0 | 0 |
meth |
Init._post_init_method |
2 | 0 | 0 |
meth |
Init._convert_to_deepspeed_param |
2 | 0 | 0 |
meth |
Init._aligned_size |
2 | 0 | 0 |
meth |
Init._padding_size |
2 | 0 | 0 |
meth |
Init._partition_numel |
2 | 0 | 0 |
meth |
Init._ensure_availability_of_partitioned_params |
2 | 0 | 0 |
meth |
Init._all_gather |
4 | 0 | 0 |
meth |
Init._partition |
5 | 0 | 0 |
meth |
Init._partition_param |
5 | 0 | 0 |
meth |
Init._partition_param_sec |
4 | 0 | 0 |
meth |
Init._param_status |
2 | 0 | 0 |
meth |
Init._allgather_param |
4 | 0 | 0 |
meth |
Init._allgather_params_coalesced |
4 | 0 | 0 |
meth |
Init._allgather_params_sequential |
3 | 0 | 0 |
meth |
Init._reduce_scatter_gradients |
2 | 0 | 0 |
meth |
Init._reduce_scatter_gradient |
2 | 0 | 0 |
meth |
Init._partition_gradients |
4 | 0 | 0 |
meth |
Init._partition_gradient |
4 | 0 | 0 |
meth |
Init.get_partition_dp_group |
2 | 0 | 0 |
meth |
Init.get_partition_rank |
1 | 0 | 0 |
meth |
Init.get_dp_process_group |
1 | 0 | 0 |
prop |
Init.num_partitions |
1 | 0 | 0 |
attr |
Init.param_persistence_threshold |
1 | 0 | 0 |
attr |
Init.model_persistence_threshold |
1 | 0 | 0 |
attr |
Init.override_module_apply |
1 | 0 | 0 |
attr |
Init.tensor_overrides |
1 | 0 | 0 |
attr |
Init.rank |
1 | 0 | 0 |
attr |
Init.dp_world_size |
1 | 0 | 0 |
attr |
Init.zero_param_process_group |
1 | 0 | 0 |
attr |
Init.num_ranks_in_param_group |
1 | 0 | 0 |
attr |
Init.rank_in_group |
1 | 0 | 0 |
attr |
Init.num_param_groups |
1 | 0 | 0 |
attr |
Init.local_device |
1 | 0 | 0 |
attr |
Init.quantized_weights |
1 | 0 | 0 |
attr |
Init.quantized_nontrainable_weights |
1 | 0 | 0 |
attr |
Init.enable_sanity_checks |
1 | 0 | 0 |
attr |
Init.module |
1 | 0 | 0 |
attr |
Init.remote_device |
1 | 0 | 0 |
attr |
Init.pin_memory |
1 | 0 | 0 |
attr |
Init.use_all_gather_into_tensor |
1 | 0 | 0 |
attr |
Init.use_all_reduce_for_fetch_params |
1 | 0 | 0 |
attr |
Init.allgather_sequential |
1 | 0 | 0 |
attr |
Init.ds_process_group |
1 | 0 | 0 |
attr |
Init.quantizer_module |
1 | 0 | 0 |
attr |
Init.param_swapper |
1 | 0 | 0 |
func |
restore_init_context |
1 | 0 | 0 |
meth |
AllGatherHandle.init |
6 | 2 | 0 |
meth |
AllGatherHandle.wait |
2 | 1 | 0 |
func |
print_rank_0 |
4 | 0 | 0 |
func |
register_external_parameter |
3 | 0 | 0 |
func |
get_all_subclasses |
3 | 0 | 0 |
meth |
NoGatherCoalescedHandle.wait |
2 | 1 | 0 |
meth |
AllGatherCoalescedHandle.init |
7 | 4 | 0 |
meth |
AllGatherCoalescedHandle.wait |
2 | 1 | 0 |
meth |
AllGatherCoalescedHandle.free_buffer |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.allgather_handle |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.params |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.partitions |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.world_size |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.use_secondary_tensor |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.complete |
1 | 0 | 0 |
attr |
AllGatherCoalescedHandle.quantization |
1 | 0 | 0 |
meth |
CUDAQuantizer.quantize |
3 | 0 | 0 |
meth |
CUDAQuantizer.dequantize |
3 | 0 | 0 |
attr |
CUDAQuantizer.group_size_cache |
1 | 0 | 0 |
func |
shutdown_init_context |
1 | 0 | 0 |
meth |
MultipleAllGatherHandles.init |
2 | 1 | 0 |
meth |
MultipleAllGatherHandles.wait |
2 | 1 | 0 |
attr |
MultipleAllGatherHandles.handles |
1 | 0 | 0 |
meth |
AllReduceCoalescedHandle.init |
3 | 2 | 0 |
meth |
AllReduceCoalescedHandle.wait |
2 | 1 | 0 |
attr |
AllReduceCoalescedHandle.handle |
1 | 0 | 0 |
attr |
AllReduceCoalescedHandle.params |
1 | 0 | 0 |
attr |
AllReduceCoalescedHandle.complete |
1 | 0 | 0 |
func |
unregister_external_parameter |
3 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses.init |
5 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses.enter |
1 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses.exit |
4 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses._post_init_method |
2 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses._set_dtype |
3 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses.patch_init_and_builtins |
1 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses.unpatch_init_and_builtins |
1 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses._add_tensor_creation_wrappers |
1 | 0 | 0 |
meth |
InsertPostInitMethodToModuleSubClasses._remove_tensor_creation_wrappers |
1 | 0 | 0 |
attr |
InsertPostInitMethodToModuleSubClasses.mem_efficient_linear |
1 | 0 | 0 |
attr |
InsertPostInitMethodToModuleSubClasses.enabled |
1 | 0 | 0 |
attr |
InsertPostInitMethodToModuleSubClasses.wrapped_cls |
1 | 0 | 0 |
attr |
InsertPostInitMethodToModuleSubClasses.skip_init_depth |
1 | 0 | 0 |
attr |
InsertPostInitMethodToModuleSubClasses.quantized_initialization |
1 | 0 | 0 |
meth |
NoGatherHandle.wait |
2 | 1 | 0 |
attr |
QuantizationInfo.quantized_param |
1 | 0 | 0 |
attr |
QuantizationInfo.backend |
1 | 0 | 0 |
attr |
QuantizationInfo.quant_handle |
1 | 0 | 0 |
attr |
QuantizationInfo.scale_buffer |
1 | 0 | 0 |
meth |
GatheredParameters.init |
5 | 0 | 0 |
meth |
GatheredParameters.enter |
1 | 0 | 0 |
meth |
GatheredParameters.exit |
2 | 0 | 0 |
attr |
GatheredParameters.enabled |
1 | 0 | 0 |
attr |
GatheredParameters.params |
1 | 0 | 0 |
attr |
GatheredParameters.enable_sanity_checks |
1 | 0 | 0 |
attr |
GatheredParameters.src_rank |
1 | 0 | 0 |
attr |
GatheredParameters.fwd_module |
1 | 0 | 0 |
deepspeed.runtime.zero.partitioned_param_coordinator (24 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
iter_params |
3 | 2 | 0 |
meth |
PartitionedParameterCoordinator.init |
12 | 7 | 0 |
meth |
PartitionedParameterCoordinator.construct_parameter_trace_from_module_trace |
1 | 0 | 0 |
meth |
PartitionedParameterCoordinator._dump_params |
5 | 0 | 0 |
meth |
PartitionedParameterCoordinator._dump_param_ids |
5 | 0 | 0 |
meth |
PartitionedParameterCoordinator.release_sub_module |
3 | 2 | 0 |
attr |
PartitionedParameterCoordinator.zero_quantized_weights |
1 | 0 | 0 |
attr |
PartitionedParameterCoordinator.zero_quantized_nontrainable_weights |
1 | 0 | 0 |
attr |
PartitionedParameterCoordinator.fast_sharding_for_leaf_module |
1 | 0 | 0 |
func |
get_all_parameters |
3 | 0 | 0 |
deepspeed.runtime.zero.partitioned_param_profiler (16 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
EventCounter.reset |
1 | 0 | 0 |
meth |
EventCounter.increment |
2 | 0 | 0 |
meth |
PartitionedParameterProfiler.init |
2 | 0 | 0 |
meth |
PartitionedParameterProfiler.reset_events |
1 | 0 | 0 |
meth |
PartitionedParameterProfiler.start_event |
2 | 0 | 0 |
meth |
PartitionedParameterProfiler.stop_event |
3 | 0 | 0 |
meth |
PartitionedParameterProfiler._log_timers |
1 | 0 | 0 |
meth |
PartitionedParameterProfiler._log_event_counters |
1 | 0 | 0 |
meth |
PartitionedParameterProfiler.log_events |
1 | 0 | 0 |
attr |
PartitionedParameterProfiler.timers |
1 | 0 | 0 |
attr |
PartitionedParameterProfiler.event_counters |
1 | 0 | 0 |
deepspeed.runtime.zero.stage3 (464 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
unwrap_model_for_generation |
2 | 0 | 0 |
func |
estimate_zero3_model_states_mem_needs_all_cold |
6 | 0 | 0 |
func |
estimate_zero3_model_states_mem_needs |
9 | 0 | 0 |
func |
estimate_zero3_model_states_mem_needs_all_live |
5 | 0 | 0 |
func |
isclose |
5 | 0 | 0 |
func |
move_to_cpu |
2 | 0 | 0 |
func |
model_to_params |
2 | 0 | 0 |
meth |
IPGBucketZ3.clear |
1 | 0 | 0 |
meth |
IPGBucketZ3.clear_params |
1 | 0 | 0 |
func |
lcm |
3 | 0 | 0 |
func |
input |
2 | 0 | 0 |
func |
print_rank_0 |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.init |
46 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.destroy |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.create_zenflow_hooks |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.initialize_ds_offload |
19 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_trainable_parameter_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_zero_group_parallelism |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.invalidate_secondary_tensor |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._setup_for_real_optimizer |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._link_all_hp_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_lr |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_lr |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_param_coordinator |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._configure_offloading |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._configure_tensor_swapping |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._move_to_flat_buffer |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._create_param_groups_fp16_flat_cpu_memory |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._create_fp16_partitions_with_defragmentation |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._swap_in_sub_group_to_flat_buffer |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._create_next_swappable_fp32_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_sub_group_partitions |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._create_fp32_partitions |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._create_fp16_sub_groups |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._optimizer_step |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._swappable_optimizer_subgroup |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._partitioned_params_swap_out |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_fp16_partitioned_groups_flat |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.initialize_optimizer_states |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_first_param_index |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.initialize_gradient_partitioning_data_structures |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.independent_gradient_partition_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.overlapping_partition_gradients_reduce_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.create_reduce_and_remove_grad_hooks |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_param_id |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.reduce_independent_p_g_buckets_and_remove_grads |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_grad_positions |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._constant_buffered_norm2 |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_norm_for_param_grad_in_gpu |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.async_inplace_copy_grad_to_fp32_buffer_from_gpu |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.complete_grad_norm_calculation_for_cpu_offload |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.reduce_ready_partitions_and_remove_grads |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.zero_reduced_gradients |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.quantize_nontrainable_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.flatten_and_print |
5 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_grads_to_reduce |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.sequential_execution |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_none_gradients_to_zero |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.allreduce_bucket |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.allreduce_and_copy |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.allreduce_no_retain |
5 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_data_parallel_partitions |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_partition_info |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.zero_grad |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.clear_backward_seen_flag |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._model_parallel_all_reduce |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_grad_norm_direct |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_flat_partition |
5 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.free_grad_in_param_list |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.reset_cpu_buffers |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._pre_step |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_norm_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._prepare_fp32_grad_for_sub_group |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._prepare_sub_group |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._optimizer_states_and_gradient_swap_in |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._release_sub_group |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.flatten_dense_tensors_aligned |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._optimizer_states_and_gradient_swap_out |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._release_swap_buffers |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._writeback_swap_state |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._unflatten_partitioned_parameters |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._overflow_clean_up |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._loco_err_buf_update |
3 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._overflow_check_and_loss_scale_update |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._post_step |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._reassign_or_swap_out_partitioned_parameters |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.override_loss_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.step |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.dump_pre_step_gradients |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.dump_post_step_gradients |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.unscale_and_clip_grads |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._check_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.has_overflow_serial |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.has_overflow_partitioned_grads_serial |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.has_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._has_inf_or_nan |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.backward_prologue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.backward_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._fp32_state_allgather |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_fp32_grad_state_partition |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_fp32_grad_for_param |
2 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_fp32_grad_for_param |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_fp32_opt_state_partition |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_full_hp_param |
3 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_full_hp_param |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_local_fp32_grad_for_param |
2 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_local_grad_for_param |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_local_fp32_param |
3 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.set_local_hp_param |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.update_fp32_grad_for_param_vectorized |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_hp_param_device |
3 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._partition_all_parameters |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.check_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._update_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_state |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_state |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_param_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_param_groups |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_loss_scale |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_loss_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_lean_tensors |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_lean_optimizer_state |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.get_groups_without_padding |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._set_fp32_optimizer_param_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._clear_fp32_optimizer_param_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._rigid_state_dict |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.state_dict |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._restore_from_fp32_weights |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._restore_from_bit16_weights |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.refresh_fp32_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._get_flattened_partition |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._restore_base_optimizer_state |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._rigid_load_state_dict |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.load_state_dict |
7 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._load_universal_checkpoint |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.load_hp_checkpoint_state_from_checkpoint_dir_stage3 |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3._load_global_state_stage3 |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.load_hp_checkpoint_state |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.reset_swap_buffers |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.checkpoint_event_prologue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.checkpoint_event_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.empty_partition_cache |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.offload_states |
5 | 4 | 0 |
meth |
DeepSpeedZeroOptimizer_Stage3.reload_states |
2 | 1 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.optimizer |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.param_names |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.flatten |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.unflatten |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.gradient_accumulation_dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.custom_loss_scaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.external_loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.optimizer_swapper |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.swap_optimizer |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.offload_optimizer |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.offload_optimizer_pin_memory |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.offload_optimizer_fast_init |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.offload_param |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.offload_param_pin_memory |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.params_in_nvme_and_cpu |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.max_params_in_cpu |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.partial_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.enable_sanity_checks |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.zero_hpz_partition_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.parameter_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.persistent_parameters |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.master_weights_and_grads_dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.module |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.elastic_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.device |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.deepspeed_adam_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.reduce_and_partition_stream |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.n_caching_allocator_flushes |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.timers |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.all2all_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.reduce_scatter |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.dp_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.sequence_parallel_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.zero_quantized_nontrainable_weights |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.partition_count |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.zeropp_loco_param |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.overflow |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.clip_grad |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.communication_data_type |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.gradient_predivide_factor |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.postscale_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.gradient_accumulation_steps |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.micro_step_id |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.reduce_bucket_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp16_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp16_partitioned_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp16_partitioned_groups_flat |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp16_partitioned_groups_flat_numel |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp16_partitioned_groups_flat_id |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.param_groups_fp16_flat_cpu_memory |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.fp32_partitioned_groups_flat |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.next_swappable_fp32_partitioned_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.partition_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.all_reduce_print |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.prefetch_elements |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.contiguous_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.groups_padding |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.sub_group_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.sub_group_to_group_id |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.trainable_param_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.param_dict |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.is_param_in_current_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.torch_autocast_gradscaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.params_already_reduced |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.previous_reduced_grads |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.grad_position |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.is_partition_reduced |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.is_grad_computed |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.averaged_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.loss_scaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.dynamic_loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.debug_fp16_grads |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.backup_optimizer |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.model_parallel_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.model_parallel_rank |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.norm_for_param_grads |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.state |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.param_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer_Stage3.cur_scale |
1 | 0 | 0 |
deepspeed.runtime.zero.stage_1_and_2 (431 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_alignment_padding |
3 | 0 | 0 |
func |
isclose |
5 | 0 | 0 |
func |
print_rank_msg |
2 | 0 | 0 |
meth |
IPGBucket.clear |
1 | 0 | 0 |
func |
estimate_zero2_model_states_mem_needs |
6 | 0 | 0 |
func |
model_to_params |
2 | 0 | 0 |
func |
split_half_float_double |
2 | 0 | 0 |
func |
estimate_zero2_model_states_mem_needs_all_live |
5 | 0 | 0 |
func |
input |
2 | 0 | 0 |
func |
lcm |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.init |
36 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.destroy |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._enable_universal_checkpoint |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._create_param_mapping |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._create_optimizer_mapping |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._link_all_hp_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._lazy_init_hp_params_optimizer_state |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.is_moe_group |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._configure_moe_settings |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._update_model_bit16_weights |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._round_robin_reorder |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._release_ipg_buffers |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.initialize_optimizer_states |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reduce_gradients |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_first_param_index |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.initialize_gradient_partitioning_data_structures |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.independent_gradient_partition_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.clear_backward_seen_flag |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reset_partition_gradient_structures |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.initialize_gradient_partition |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.overlapping_partition_gradients_reduce_epilogue |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._fill_param_grad_accum_attribute |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.fill_grad_accum_attribute |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_gradient_for_reduction |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_param_gradient_attribute |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.clear_grad_attribute |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.create_gradient_handling_hooks |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_param_id |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.flatten_dense_tensors_aligned |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reduce_independent_p_g_buckets_and_remove_grads |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.print_rank_0 |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.gradient_reduction_w_predivide |
3 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.allreduce_and_copy_with_multiple_ranks |
7 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.allreduce_and_scatter |
7 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.average_tensor |
3 | 2 | 0 |
meth |
DeepSpeedZeroOptimizer.get_grad_position |
5 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.update_offload_overflow_tracker |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.update_offload_overflow_tracker_for_param_grad |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_offload_gradient_dict |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.async_accumulate_grad_in_cpu_via_gpu |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.set_norm_for_param_grad |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.set_norm_for_param_grad_in_gpu |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.async_inplace_copy_grad_to_fp32_buffer_from_gpu |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.complete_grad_norm_calculation_for_cpu_offload |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.copy_grads_in_partition |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reduce_ipg_grads |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.process_gradients |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reduce_ready_partitions_and_remove_grads |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.zero_reduced_gradients |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.flatten_and_print |
5 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_grads_to_reduce |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.sequential_execution |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.set_none_gradients_to_zero |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.allreduce_bucket |
7 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer._clear_previous_reduced_grads |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.allreduce_and_copy |
7 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.allreduce_no_retain |
8 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.buffered_reduce_fallback |
6 | 1 | 0 |
meth |
DeepSpeedZeroOptimizer.get_data_parallel_partitions |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_partition_info |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.zero_grad |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._clear_param_grad_only |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._model_parallel_all_reduce |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_grad_norm_direct |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_all_grad_tensors |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_flat_partition |
8 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.free_grad_in_param_list |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.reset_cpu_buffers |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.set_lr |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_lr |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.override_loss_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.scaled_global_norm |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_bit16_param_group |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._optimizer_step |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.step |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.update_lp_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._average_expert_grad_norms |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.unscale_and_clip_grads |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._check_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.has_overflow_serial |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.has_overflow_partitioned_grads_serial |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.has_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._has_inf_or_nan |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.setup_buckets |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.backward_epilogue |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.check_overflow |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._update_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_state |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._set_state |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_param_groups |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._set_param_groups |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_loss_scale |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._set_loss_scale |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_groups_without_padding |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_state_without_padding |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._get_base_optimizer_state |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.state_dict |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._restore_from_elastic_fp32_weights |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._restore_from_bit16_weights |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.refresh_fp32_params |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._partition_base_optimizer_state |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._restore_step_from_elastic_checkpoint |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._restore_base_optimizer_state |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.get_ep_ranks |
3 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._restore_elastic_base_optimizer_state |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.load_state_dict |
7 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._load_universal_checkpoint |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._load_global_state |
2 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._load_legacy_checkpoint |
4 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._clear_hp_buffer_references |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer._clear_lp_params_references |
1 | 0 | 0 |
meth |
DeepSpeedZeroOptimizer.offload_states |
5 | 4 | 0 |
meth |
DeepSpeedZeroOptimizer.reload_states |
2 | 1 | 0 |
attr |
DeepSpeedZeroOptimizer.zenflow |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.elastic_checkpoint |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.check_grad_overflow |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_names |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.mpu |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.optimizer |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.flatten |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.unflatten |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.partition_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.zero_stage_string |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.timers |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.reduce_scatter |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.overlap_comm |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.deepspeed_adam_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.device |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.dp_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.sequence_parallel_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.ep_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.expert_dp_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.real_dp_process_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.partition_count |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.is_gradient_accumulation_boundary |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.contiguous_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.has_moe_layers |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.overflow |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.clip_grad |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.communication_data_type |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.gradient_predivide_factor |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.postscale_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.gradient_accumulation_steps |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.micro_step_id |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.ignore_unused_parameters |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.round_robin_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.master_weights_and_grads_dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.low_precision_master_weights_and_grads |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.bit16_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.bit16_groups_flat |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.parallel_partitioned_bit16_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.single_partition_of_fp32_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.params_not_in_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.params_in_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.first_offset |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.partition_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.nccl_start_alignment_factor |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.all_reduce_print |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.gradient_accumulation_dtype |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.round_robin_bit16_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.round_robin_bit16_indices |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.round_robin_bit16_meta |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.groups_padding |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.reduce_bucket_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.use_multi_rank_bucket_allreduce |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.allgather_bucket_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.reduction_stream |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.callback_queued |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_dict |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.is_param_in_current_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.torch_autocast_gradscaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.params_already_reduced |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_id |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_to_partition_ids |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.is_partition_reduced |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.remaining_grads_in_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.total_grads_in_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.is_grad_computed |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.grad_partition_insertion_offset |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.grad_start_offset |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.averaged_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.all_grad_tensors |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.offload_gradient_dict |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.first_param_index_in_partition |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.ready_for_gradients |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.custom_loss_scaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.external_loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.loss_scaler |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.dynamic_loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.cpu_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.cpu_offload_pin_memory |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.model_parallel_group |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.model_parallel_world_size |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.model_parallel_rank |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_buffer_of_bit16_for_cpu_offload_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.use_separate_grad_accum |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.use_grad_accum_attribute |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.accumulated_grads_in_cpu |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.norm_for_param_grads |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.local_overflow |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.grad_position |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.temp_grad_buffer_for_cpu_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.temp_grad_buffer_for_gpu_offload |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.state |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.param_groups |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.loss_scale |
1 | 0 | 0 |
attr |
DeepSpeedZeroOptimizer.cur_scale |
1 | 0 | 0 |
func |
estimate_zero2_model_states_mem_needs_all_cold |
5 | 0 | 0 |
deepspeed.runtime.zero.test (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
test1 |
1 | 0 | 0 |
func |
test2 |
1 | 0 | 0 |
deepspeed.runtime.zero.tiling (46 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
TiledLinear.init |
11 | 0 | 0 |
meth |
TiledLinear.forward |
2 | 0 | 0 |
meth |
TiledLinear._split_global_input |
3 | 0 | 0 |
meth |
TiledLinear._reduce_local_output |
5 | 0 | 0 |
meth |
TiledLinear._combine_output_splits |
2 | 0 | 0 |
meth |
TiledLinear.copy_params_from |
2 | 0 | 0 |
attr |
TiledLinear.in_features |
1 | 0 | 0 |
attr |
TiledLinear.out_features |
1 | 0 | 0 |
attr |
TiledLinear.use_bias |
1 | 0 | 0 |
attr |
TiledLinear.out_splits |
1 | 0 | 0 |
attr |
TiledLinear.in_splits |
1 | 0 | 0 |
attr |
TiledLinear.input_is_already_split |
1 | 0 | 0 |
attr |
TiledLinear.combine_out_splits |
1 | 0 | 0 |
attr |
TiledLinear.in_parts |
1 | 0 | 0 |
attr |
TiledLinear.out_parts |
1 | 0 | 0 |
attr |
TiledLinear.linears |
1 | 0 | 0 |
meth |
TiledLinearReturnBias._reduce_local_output |
5 | 0 | 0 |
meth |
TiledLinearReturnBias._combine_output_splits |
2 | 0 | 0 |
func |
split_tensor_along_last_dim |
4 | 0 | 0 |
deepspeed.runtime.zero.utils (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
apply_to_tensors_only |
4 | 0 | 0 |
func |
is_builtin_type |
2 | 0 | 0 |
func |
is_zero_param |
2 | 0 | 0 |
func |
is_zero_supported_optimizer |
2 | 0 | 0 |
deepspeed.sequence.cross_entropy (4 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
vocab_sequence_parallel_cross_entropy |
4 | 0 | 0 |
deepspeed.sequence.fpdt_layer (99 missing, 2 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
FPDT_InputConstruct.init |
9 | 1 | 0 |
meth |
FPDT_InputConstruct.generate |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.tokens |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.labels |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.loss_mask |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.attention_mask |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.position_ids |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.num_chunk_per_gpu |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.chunk_size |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.sp_size |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.sp_rank |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.global_seq_len |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.local_seq_len |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.batch_size |
1 | 0 | 0 |
attr |
FPDT_InputConstruct.device |
1 | 0 | 0 |
meth |
SequenceChunk.init |
4 | 1 | 0 |
meth |
SequenceChunk.load_to_gpu |
1 | 0 | 0 |
meth |
SequenceChunk.get_gpu_chunk |
1 | 0 | 0 |
meth |
SequenceChunk.check_gpu_chunk |
1 | 0 | 0 |
meth |
SequenceChunk.offload |
1 | 0 | 0 |
meth |
SequenceChunk.overwrite_to_cpu |
1 | 0 | 0 |
attr |
SequenceChunk.chunk_shape |
1 | 0 | 0 |
attr |
SequenceChunk.chunk_dtype |
1 | 0 | 0 |
attr |
SequenceChunk.device |
1 | 0 | 0 |
attr |
SequenceChunk.cpu_chunk |
1 | 0 | 0 |
attr |
SequenceChunk.gpu_chunk |
1 | 0 | 0 |
meth |
FPDT_LogitsLoss.forward |
9 | 1 | 1 |
meth |
FPDT_LogitsLoss.backward |
3 | 0 | 0 |
func |
update_out_and_lse |
6 | 5 | 0 |
func |
apply_rotary_pos_emb_backward |
4 | 0 | 0 |
func |
bias_gelu |
2 | 0 | 0 |
meth |
FPDT_FFN.forward |
9 | 1 | 1 |
meth |
FPDT_FFN.backward |
4 | 0 | 0 |
meth |
FPDT_Attention.init |
12 | 3 | 0 |
meth |
FPDT_Attention.forward |
6 | 1 | 0 |
attr |
FPDT_Attention.spg |
1 | 0 | 0 |
attr |
FPDT_Attention.scatter_idx |
1 | 0 | 0 |
attr |
FPDT_Attention.gather_idx |
1 | 0 | 0 |
attr |
FPDT_Attention.config |
1 | 0 | 0 |
attr |
FPDT_Attention.projection_size |
1 | 0 | 0 |
attr |
FPDT_Attention.hidden_size_per_attention_head |
1 | 0 | 0 |
attr |
FPDT_Attention.kv_projection_size |
1 | 0 | 0 |
attr |
FPDT_Attention.hidden_size |
1 | 0 | 0 |
attr |
FPDT_Attention.qkv_linear_weight |
1 | 0 | 0 |
attr |
FPDT_Attention.qkv_linear_bias |
1 | 0 | 0 |
attr |
FPDT_Attention.qkv_dense_weight |
1 | 0 | 0 |
attr |
FPDT_Attention.qkv_dense_bias |
1 | 0 | 0 |
attr |
FPDT_Attention.reture_bias |
1 | 0 | 0 |
attr |
FPDT_Attention.dropout |
1 | 0 | 0 |
attr |
FPDT_Attention.chunk_size |
1 | 0 | 0 |
attr |
FPDT_Attention.double_buffer |
1 | 0 | 0 |
func |
bias_gelu_back |
3 | 0 | 0 |
attr |
flash_attn_version |
1 | 0 | 0 |
deepspeed.sequence.layer (39 missing, 1 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
pre_all2all_fun |
4 | 0 | 0 |
func |
single_all_to_all |
9 | 0 | 0 |
meth |
DistributedAttention.init |
6 | 5 | 0 |
meth |
DistributedAttention.layer_sync |
2 | 0 | 0 |
meth |
DistributedAttention.forward |
8 | 6 | 1 |
attr |
DistributedAttention.local_attn |
1 | 0 | 0 |
attr |
DistributedAttention.spg |
1 | 0 | 0 |
attr |
DistributedAttention.scatter_idx |
1 | 0 | 0 |
attr |
DistributedAttention.gather_idx |
1 | 0 | 0 |
attr |
DistributedAttention.sp_overlap_comm |
1 | 0 | 0 |
attr |
DistributedAttention.overlap_handles |
1 | 0 | 0 |
attr |
DistributedAttention.sp_stream |
1 | 0 | 0 |
attr |
DistributedAttention.default_stream |
1 | 0 | 0 |
func |
uneven_heads_all2all |
6 | 0 | 0 |
func |
post_all2all |
3 | 0 | 0 |
func |
apply_rotary_pos_emb |
4 | 0 | 0 |
deepspeed.utils.bwc (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
bwc_tensor_model_parallel_group |
2 | 0 | 0 |
func |
bwc_pipeline_parallel_world_size |
2 | 0 | 0 |
func |
bwc_tensor_model_parallel_world_size |
2 | 0 | 0 |
func |
bwc_tensor_model_parallel_rank |
2 | 0 | 0 |
func |
bwc_pipeline_parallel_group |
2 | 0 | 0 |
deepspeed.utils.comms_logging (41 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
convert_size |
2 | 0 | 0 |
meth |
CommsLogger.init |
1 | 0 | 0 |
meth |
CommsLogger.configure |
2 | 0 | 0 |
meth |
CommsLogger.start_profiling_comms |
1 | 0 | 0 |
meth |
CommsLogger.stop_profiling_comms |
1 | 0 | 0 |
meth |
CommsLogger.start_profiling_op |
2 | 0 | 0 |
meth |
CommsLogger.stop_profiling_op |
2 | 0 | 0 |
meth |
CommsLogger.append |
5 | 0 | 0 |
meth |
CommsLogger.get_raw_data |
1 | 0 | 0 |
meth |
CommsLogger.has_data |
1 | 0 | 0 |
meth |
CommsLogger.reset_data |
1 | 0 | 0 |
meth |
CommsLogger.get_operation_names |
1 | 0 | 0 |
meth |
CommsLogger.get_total_operations |
1 | 0 | 0 |
meth |
CommsLogger.get_operation_summary |
2 | 0 | 0 |
meth |
CommsLogger.log_all |
4 | 0 | 0 |
attr |
CommsLogger.comms_dict |
1 | 0 | 0 |
attr |
CommsLogger.verbose |
1 | 0 | 0 |
attr |
CommsLogger.debug |
1 | 0 | 0 |
attr |
CommsLogger.prof_ops |
1 | 0 | 0 |
attr |
CommsLogger.prof_all |
1 | 0 | 0 |
attr |
CommsLogger.enabled |
1 | 0 | 0 |
func |
get_caller_func |
2 | 0 | 0 |
func |
calc_bw_log |
4 | 0 | 0 |
func |
print_rank_0 |
2 | 0 | 0 |
deepspeed.utils.config (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_timers_config |
2 | 0 | 0 |
deepspeed.utils.debug (38 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
debug_module2name_id |
2 | 0 | 0 |
func |
debug_param2name_id_shape_status |
2 | 0 | 0 |
func |
debug_param2name |
2 | 0 | 0 |
func |
debug_module2name |
2 | 0 | 0 |
func |
ds_shape |
2 | 0 | 0 |
func |
debug_param2name_id_shape_device |
2 | 0 | 0 |
func |
print_rank0 |
3 | 0 | 0 |
func |
printflock |
2 | 0 | 0 |
func |
ds_id |
2 | 0 | 0 |
func |
log_rank_file |
3 | 0 | 0 |
func |
print_backward_tensors |
2 | 0 | 0 |
func |
debug_param2name_id_numel |
2 | 0 | 0 |
func |
debug_clear_module_and_param_names |
1 | 0 | 0 |
func |
print_rank |
3 | 0 | 0 |
func |
debug_module2name_class |
2 | 0 | 0 |
func |
debug_param2name_id |
2 | 0 | 0 |
func |
debug_param2name_id_shape |
2 | 0 | 0 |
func |
debug_extract_module_and_param_names |
2 | 0 | 0 |
deepspeed.utils.groups (17 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
get_tensor_model_parallel_src_rank |
1 | 0 | 0 |
func |
get_model_parallel_world_size |
1 | 0 | 0 |
func |
get_tensor_model_parallel_group |
1 | 0 | 0 |
func |
get_model_parallel_rank |
1 | 0 | 0 |
func |
set_tensor_model_parallel_world_size |
2 | 0 | 0 |
func |
get_tensor_model_parallel_world_size |
1 | 0 | 0 |
func |
get_data_parallel_world_size |
1 | 0 | 0 |
func |
get_tensor_model_parallel_rank |
1 | 0 | 0 |
func |
get_data_parallel_rank |
1 | 0 | 0 |
func |
get_model_parallel_group |
1 | 0 | 0 |
func |
get_data_parallel_group |
1 | 0 | 0 |
func |
set_tensor_model_parallel_rank |
2 | 0 | 0 |
func |
initialize |
3 | 0 | 0 |
deepspeed.utils.init_on_device (12 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
OnDevice.init |
4 | 0 | 0 |
meth |
OnDevice.enter |
1 | 0 | 0 |
meth |
OnDevice.exit |
4 | 0 | 0 |
attr |
OnDevice.dtype |
1 | 0 | 0 |
attr |
OnDevice.enabled |
1 | 0 | 0 |
attr |
OnDevice.device |
1 | 0 | 0 |
deepspeed.utils.logging (36 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
warning_once |
3 | 0 | 0 |
func |
log_dist_once |
4 | 0 | 0 |
func |
set_log_level_from_string |
3 | 0 | 0 |
attr |
logger |
1 | 0 | 0 |
func |
print_configuration |
3 | 0 | 0 |
func |
print_dist |
3 | 0 | 0 |
func |
get_dist_msg |
3 | 0 | 0 |
func |
log_dist |
4 | 0 | 0 |
func |
get_current_level |
1 | 0 | 0 |
func |
print_json_dist |
4 | 0 | 0 |
func |
get_log_level_from_string |
2 | 0 | 0 |
meth |
LoggerFactory.create_logger |
3 | 0 | 0 |
func |
should_log_le |
2 | 0 | 0 |
deepspeed.utils.mixed_precision_linkage (14 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
link_hp_params |
10 | 0 | 0 |
func |
lazy_init_hp_params_optimizer_state |
4 | 0 | 0 |
deepspeed.utils.numa (10 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
parse_range_list |
2 | 0 | 0 |
func |
parse_range |
2 | 0 | 0 |
func |
get_numa_cores |
1 | 0 | 0 |
func |
get_numactl_cmd |
4 | 0 | 0 |
func |
check_for_numactl_pkg |
1 | 0 | 0 |
deepspeed.utils.nvtx (2 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
instrument_w_nvtx |
2 | 0 | 0 |
deepspeed.utils.tensor_fragment (75 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
safe_get_local_fp32_param |
2 | 0 | 0 |
meth |
tensor_fragment.update_hp |
1 | 0 | 0 |
meth |
tensor_fragment.update_lp |
1 | 0 | 0 |
meth |
tensor_fragment.get_optim_state_fragment |
2 | 0 | 0 |
meth |
tensor_fragment.set_optim_state_fragment |
3 | 0 | 0 |
meth |
tensor_fragment.get_hp_fragment_address |
1 | 0 | 0 |
meth |
tensor_fragment.get_optim_state_keys |
1 | 0 | 0 |
meth |
tensor_fragment.get_hp_fragment |
2 | 0 | 0 |
meth |
tensor_fragment.get_lp_grad_fragment |
2 | 0 | 0 |
func |
map_to_flat_opt_states |
5 | 0 | 0 |
func |
safe_get_full_fp32_param |
2 | 0 | 0 |
func |
get_hp_fragment_mapping |
10 | 0 | 0 |
func |
safe_set_local_optimizer_state |
4 | 0 | 0 |
func |
get_full_hp_grad |
2 | 0 | 0 |
func |
safe_update_full_grad_vectorized |
3 | 2 | 0 |
func |
safe_set_local_grad |
3 | 0 | 0 |
func |
get_full_hp_param |
3 | 0 | 0 |
func |
set_full_hp_param |
4 | 0 | 0 |
func |
safe_set_full_optimizer_state |
4 | 0 | 0 |
func |
safe_get_local_optimizer_state |
3 | 0 | 0 |
func |
safe_set_local_fp32_param |
3 | 0 | 0 |
func |
safe_get_full_grad |
2 | 0 | 0 |
func |
safe_get_local_grad |
2 | 0 | 0 |
func |
safe_set_full_fp32_param |
3 | 0 | 0 |
func |
safe_get_full_optimizer_state |
3 | 0 | 0 |
func |
safe_set_full_grad |
3 | 0 | 0 |
func |
set_full_hp_grad |
3 | 0 | 0 |
deepspeed.utils.timer (87 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
meth |
SynchronizedWallClockTimer.init |
1 | 0 | 0 |
meth |
SynchronizedWallClockTimer.get_timers |
1 | 0 | 0 |
meth |
SynchronizedWallClockTimer.call |
2 | 0 | 0 |
meth |
SynchronizedWallClockTimer.memory_usage |
1 | 0 | 0 |
meth |
SynchronizedWallClockTimer.log |
6 | 0 | 0 |
meth |
SynchronizedWallClockTimer.get_mean |
4 | 0 | 0 |
attr |
SynchronizedWallClockTimer.timers |
1 | 0 | 0 |
meth |
ThroughputTimer.init |
7 | 0 | 0 |
meth |
ThroughputTimer.update_epoch_count |
1 | 0 | 0 |
meth |
ThroughputTimer.init_timer |
1 | 0 | 0 |
meth |
ThroughputTimer.start |
1 | 0 | 0 |
meth |
ThroughputTimer._is_report_boundary |
1 | 0 | 0 |
meth |
ThroughputTimer.stop |
3 | 0 | 0 |
meth |
ThroughputTimer.avg_samples_per_sec |
1 | 0 | 0 |
attr |
ThroughputTimer.config |
1 | 0 | 0 |
attr |
ThroughputTimer.start_time |
1 | 0 | 0 |
attr |
ThroughputTimer.end_time |
1 | 0 | 0 |
attr |
ThroughputTimer.started |
1 | 0 | 0 |
attr |
ThroughputTimer.batch_size |
1 | 0 | 0 |
attr |
ThroughputTimer.start_step |
1 | 0 | 0 |
attr |
ThroughputTimer.epoch_count |
1 | 0 | 0 |
attr |
ThroughputTimer.micro_step_count |
1 | 0 | 0 |
attr |
ThroughputTimer.global_step_count |
1 | 0 | 0 |
attr |
ThroughputTimer.total_elapsed_time |
1 | 0 | 0 |
attr |
ThroughputTimer.step_elapsed_time |
1 | 0 | 0 |
attr |
ThroughputTimer.steps_per_output |
1 | 0 | 0 |
attr |
ThroughputTimer.monitor_memory |
1 | 0 | 0 |
attr |
ThroughputTimer.logging |
1 | 0 | 0 |
attr |
ThroughputTimer.initialized |
1 | 0 | 0 |
meth |
NoopTimer.__init__ |
1 | 0 | 0 |
meth |
NoopTimer.__call__ |
2 | 0 | 0 |
meth |
NoopTimer.get_timers |
1 | 0 | 0 |
meth |
NoopTimer.log |
6 | 0 | 0 |
meth |
NoopTimer.get_mean |
4 | 0 | 0 |
attr |
NoopTimer.timer |
1 | 0 | 0 |
func |
trim_mean |
3 | 0 | 0 |
meth |
Timer.__init__ |
2 | 0 | 0 |
meth |
Timer.start |
1 | 0 | 0 |
meth |
Timer.stop |
3 | 0 | 0 |
meth |
Timer._get_elapsed_msec |
1 | 0 | 0 |
meth |
Timer.reset |
1 | 0 | 0 |
meth |
Timer.elapsed |
2 | 0 | 0 |
meth |
Timer.mean |
1 | 0 | 0 |
attr |
Timer.name |
1 | 0 | 0 |
attr |
Timer.started_ |
1 | 0 | 0 |
attr |
Timer.event_timers |
1 | 0 | 0 |
attr |
Timer.use_host_timer |
1 | 0 | 0 |
attr |
Timer.start_event |
1 | 0 | 0 |
attr |
Timer.elapsed_records |
1 | 0 | 0 |
attr |
Timer.start_time |
1 | 0 | 0 |
attr |
Timer.end_time |
1 | 0 | 0 |
meth |
CudaEventTimer.init |
3 | 2 | 0 |
meth |
CudaEventTimer.get_elapsed_msec |
1 | 0 | 0 |
attr |
CudaEventTimer.start_event |
1 | 0 | 0 |
attr |
CudaEventTimer.end_event |
1 | 0 | 0 |
deepspeed.utils.torch (8 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
register_grad_hook |
3 | 0 | 0 |
func |
required_torch_version |
3 | 0 | 0 |
func |
jit_script_compat |
2 | 0 | 0 |
deepspeed.utils.z3_leaf_module (1 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
func |
set_z3_leaf_module |
3 | 2 | 0 |
deepspeed.utils.zero_to_fp32 (57 missing, 0 any)
| Symbol | Typable | Typed | Any | |
|---|---|---|---|---|
attr |
args |
1 | 0 | 0 |
attr |
device |
1 | 0 | 0 |
func |
parse_optim_states |
3 | 0 | 0 |
func |
zero3_partitioned_param_info |
3 | 0 | 0 |
func |
get_fp32_state_dict_from_zero_checkpoint |
5 | 0 | 0 |
func |
to_torch_tensor |
3 | 0 | 0 |
func |
parse_model_states |
2 | 0 | 0 |
func |
get_checkpoint_files |
3 | 0 | 0 |
func |
atoi |
2 | 0 | 0 |
attr |
parser |
1 | 0 | 0 |
func |
get_optim_files |
2 | 0 | 0 |
func |
get_model_state_files |
2 | 0 | 0 |
func |
natural_keys |
2 | 0 | 0 |
func |
load_state_dict_from_zero_checkpoint |
4 | 0 | 0 |
meth |
GatheredTensor.init |
6 | 0 | 0 |
meth |
GatheredTensor.contiguous |
1 | 0 | 0 |
attr |
GatheredTensor.flat_groups |
1 | 0 | 0 |
attr |
GatheredTensor.flat_groups_offset |
1 | 0 | 0 |
attr |
GatheredTensor.offset |
1 | 0 | 0 |
attr |
GatheredTensor.partitioned_numel |
1 | 0 | 0 |
attr |
GatheredTensor.shape |
1 | 0 | 0 |
attr |
GatheredTensor.dtype |
1 | 0 | 0 |
func |
get_model_state_file |
3 | 0 | 0 |
func |
convert_zero_checkpoint_to_fp32_state_dict |
7 | 0 | 0 |
Type-Ignore Comments
| Flavor | Count |
|---|---|
type: ignore |
15 |